July 2012 - cluster-commits - Fedora mailing-lists

cluster: RHEL59 - rgmanager: Exit uncleanly only when CMAN_SHUTDOWN_ANYWAY is set
by Ryan McCabe 27 Jul '12

27 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=0c5ff64499bbb… Commit: 0c5ff64499bbb510a83afab6a7180185e08060f9 Parent: 01d917b0f2bb773ae62c2eb320e9877647803b70 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Fri Jul 27 13:03:04 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Fri Jul 27 15:54:51 2012 -0400 rgmanager: Exit uncleanly only when CMAN_SHUTDOWN_ANYWAY is set Only exit uncleanly when the CMAN_SHUTDOWN_ANYWAY flag is set in the argument passed when handling the CMAN_REASON_TRY_SHUTDOWN event. This fixes the case where args is 2, where we want to refuse to shut down. Resolves: rhbz#769730 Acked-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/clulib/msg_cluster.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/rgmanager/src/clulib/msg_cluster.c b/rgmanager/src/clulib/msg_cluster.c index e864853..e4b6b39 100644 --- a/rgmanager/src/clulib/msg_cluster.c +++ b/rgmanager/src/clulib/msg_cluster.c @@ -211,7 +211,7 @@ poll_cluster_messages(int timeout) if (cman_dispatch(ch, 0) < 0) { process_cman_event(ch, NULL, - CMAN_REASON_TRY_SHUTDOWN, 1); + CMAN_REASON_TRY_SHUTDOWN, CMAN_SHUTDOWN_ANYWAY); } ret = 0; } @@ -987,7 +987,9 @@ process_cman_event(cman_handle_t handle, void *private, int reason, int arg) printf("EVENT: %p %p %d %d\n", handle, private, reason, arg); #endif - if (reason == CMAN_REASON_TRY_SHUTDOWN && !arg) { + if (reason == CMAN_REASON_TRY_SHUTDOWN && + !(arg & CMAN_SHUTDOWN_ANYWAY)) + { cman_replyto_shutdown(handle, 0); return; }

1 0

cluster: RHEL59 - fence_scsi: fix typos in debug messages
by Fabio M. Di Nitto 27 Jul '12

27 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=01d917b0f2bb7… Commit: 01d917b0f2bb773ae62c2eb320e9877647803b70 Parent: 0a588f475afdb78660b424587f2c6460d4f1be82 Author: Ryan O'Hara <rohara(a)redhat.com> AuthorDate: Tue Apr 17 18:40:31 2012 -0500 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Fri Jul 27 09:35:25 2012 +0200 fence_scsi: fix typos in debug messages Resolves: rhbz#674497 Signed-off-by: Ryan O'Hara <rohara(a)redhat.com> Reviewed-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- fence/agents/scsi/fence_scsi.pl | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/fence/agents/scsi/fence_scsi.pl b/fence/agents/scsi/fence_scsi.pl index 91f113d..84cee91 100755 --- a/fence/agents/scsi/fence_scsi.pl +++ b/fence/agents/scsi/fence_scsi.pl @@ -111,7 +111,7 @@ sub get_node_id sub get_node_name { - print "[$pname]: get_hode_name = $opt_n\n" if $opt_v; + print "[$pname]: get_node_name = $opt_n\n" if $opt_v; return $opt_n; } @@ -163,7 +163,7 @@ sub get_host_name } } - print "[$pname]: get_host_nam = $host_name\n" if $opt_v; + print "[$pname]: get_host_name = $host_name\n" if $opt_v; return $host_name; }

1 0

fence-agents: the annotated tag v3.1.9 has been created
by Marek Grác 27 Jul '12

27 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=9ca5b914… Commit: 9ca5b9143900968bab55104a41b8074758a62181 Parent: 0000000000000000000000000000000000000000 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: 2012-07-27 06:50 +0000 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: 2012-07-27 06:50 +0000 annotated tag: v3.1.9 has been created at 9ca5b9143900968bab55104a41b8074758a62181 (tag) tagging cd73f094f7c163aca57a1dc9d5138fb468846835 (commit) replaces v3.1.8 v3.1.9 release Fabio M. Di Nitto (2): test commit build: fix dist_man_* vs man_* Lon Hohberger (1): Add iPDU fencing agent for model 46M4002 Marek 'marx' Grac (8): fence_brocade: support option action on STDIN, originally only operation was supported fence agents: Fix unique attribute for agents which are not based on fencing library fence agents: autodetect of EOL in fence agents fence_hpblade: Fence agent for HP BladeSystem fence agents: Some agents do not support action=metadata on STDIN fence_ipdu: Minor fixes to fence agent Add fence agents specific for HP iLO2, iLO3, IMM and iDrac fence_ilo2 was not correctly cleaned by Makefile

1 0

cluster: RHEL6 - qdiskd: restrict master_wins to 2 node cluster
by Fabio M. Di Nitto 27 Jul '12

27 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=4b893d2a9f46d… Commit: 4b893d2a9f46d643859fa50856ac63e0cdbd4a02 Parent: 982f2aa377fd1c0471b714945c401fdfe86e9bc1 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Mon Jul 9 08:33:37 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Fri Jul 27 08:47:10 2012 +0200 qdiskd: restrict master_wins to 2 node cluster given enough mingling of cluster.conf it was possible to break quorum rule #1: there is only one quorum in a cluster at any given time. this change restricts master_wins to 2 node cluster only and provides extra feedback to the user (via logging) on why the mode is disabled. Resolves: rhbz#838047 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Lon Hohberger <lhh(a)redhat.com> --- cman/man/qdisk.5 | 5 +++-- cman/qdisk/disk.h | 1 + cman/qdisk/main.c | 22 +++++++++++++++------- 3 files changed, 19 insertions(+), 9 deletions(-) diff --git a/cman/man/qdisk.5 b/cman/man/qdisk.5 index ca974fa..938ed69 100644 --- a/cman/man/qdisk.5 +++ b/cman/man/qdisk.5 @@ -297,8 +297,9 @@ and qdiskd's timeout (interval*tko) should be less than half of Totem's token timeout. See section 3.3.1 for more information. This option only takes effect if there are no heuristics -configured. Usage of this option in configurations with more than -two cluster nodes is undefined and should not be done. +configured and it is valid only for 2 node cluster. +This option is automatically disabled if heuristics are +defined or cluster has more than 2 nodes configured. In a two-node cluster with no heuristics and no defined vote count (see above), this mode is turned by default. If enabled in diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index 6bed41d..1d8f7c8 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -252,6 +252,7 @@ typedef struct { int qc_master; /* Master?! */ int qc_config; int qc_token_timeout; + int qc_auto_votes; disk_node_state_t qc_disk_status; disk_node_state_t qc_status; run_flag_t qc_flags; diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index 594e9e8..47ef5d2 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -1495,7 +1495,7 @@ auto_qdisk_votes(int desc) logt_print(LOG_ERR, "Unable to determine qdiskd votes " "automatically\n"); else - logt_print(LOG_DEBUG, "Setting votes to %d\n", ret); + logt_print(LOG_DEBUG, "Setting autocalculated votes to %d\n", ret); return (ret); } @@ -1657,6 +1657,8 @@ get_dynamic_config_data(qd_ctx *ctx, int ccsfd) ctx->qc_flags &= ~RF_AUTO_VOTES; } + ctx->qc_auto_votes = auto_qdisk_votes(ccsfd); + snprintf(query, sizeof(query), "/cluster/quorumd/@votes"); if (ccs_get(ccsfd, query, &val) == 0) { ctx->qc_votes = atoi(val); @@ -1664,7 +1666,7 @@ get_dynamic_config_data(qd_ctx *ctx, int ccsfd) if (ctx->qc_votes < 0) ctx->qc_votes = 0; } else { - ctx->qc_votes = auto_qdisk_votes(ccsfd); + ctx->qc_votes = ctx->qc_auto_votes; if (ctx->qc_votes < 0) { if (ctx->qc_config) { logt_print(LOG_WARNING, "Unable to determine " @@ -1930,15 +1932,21 @@ get_config_data(qd_ctx *ctx, struct h_data *h, int maxh, int *cfh) *cfh = configure_heuristics(ccsfd, h, maxh, ctx->qc_interval * (ctx->qc_tko - 1)); - if (*cfh) { - if (ctx->qc_flags & RF_MASTER_WINS) { - logt_print(LOG_WARNING, "Master-wins mode disabled\n"); + if (ctx->qc_flags & RF_MASTER_WINS) { + if (*cfh) { + logt_print(LOG_WARNING, "Master-wins mode disabled " + "(not compatible with heuristics)\n"); + ctx->qc_flags &= ~RF_MASTER_WINS; + } + if (ctx->qc_auto_votes != 1) { + logt_print(LOG_WARNING, "Master-wins mode disabled " + "(not compatible with more than 2 nodes)\n"); ctx->qc_flags &= ~RF_MASTER_WINS; } } else { if (ctx->qc_flags & RF_AUTO_VOTES && - !(ctx->qc_flags & RF_MASTER_WINS) && - ctx->qc_votes == 1) { + !*cfh && + ctx->qc_auto_votes == 1) { /* Two node cluster, no heuristics, 1 vote for * quorum disk daemon. Safe to enable master-wins. * In fact, qdiskd without master-wins in this config

1 0

cluster: STABLE32 - qdiskd: restrict master_wins to 2 node cluster
by Fabio M. Di Nitto 27 Jul '12

27 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=ebac037f4fa11… Commit: ebac037f4fa11de5f803c6025c5b182c962e84d1 Parent: c65a13a82189f38bd80d5c52924d09a7d98daa37 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Mon Jul 9 08:33:37 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Fri Jul 27 08:45:37 2012 +0200 qdiskd: restrict master_wins to 2 node cluster given enough mingling of cluster.conf it was possible to break quorum rule #1: there is only one quorum in a cluster at any given time. this change restricts master_wins to 2 node cluster only and provides extra feedback to the user (via logging) on why the mode is disabled. Resolves: rhbz#838047 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Lon Hohberger <lhh(a)redhat.com> --- cman/man/qdisk.5 | 5 +++-- cman/qdisk/disk.h | 1 + cman/qdisk/main.c | 22 +++++++++++++++------- 3 files changed, 19 insertions(+), 9 deletions(-) diff --git a/cman/man/qdisk.5 b/cman/man/qdisk.5 index ca974fa..938ed69 100644 --- a/cman/man/qdisk.5 +++ b/cman/man/qdisk.5 @@ -297,8 +297,9 @@ and qdiskd's timeout (interval*tko) should be less than half of Totem's token timeout. See section 3.3.1 for more information. This option only takes effect if there are no heuristics -configured. Usage of this option in configurations with more than -two cluster nodes is undefined and should not be done. +configured and it is valid only for 2 node cluster. +This option is automatically disabled if heuristics are +defined or cluster has more than 2 nodes configured. In a two-node cluster with no heuristics and no defined vote count (see above), this mode is turned by default. If enabled in diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index 6bed41d..1d8f7c8 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -252,6 +252,7 @@ typedef struct { int qc_master; /* Master?! */ int qc_config; int qc_token_timeout; + int qc_auto_votes; disk_node_state_t qc_disk_status; disk_node_state_t qc_status; run_flag_t qc_flags; diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index 16c26e4..8eb9a3a 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -1505,7 +1505,7 @@ auto_qdisk_votes(int desc) logt_print(LOG_ERR, "Unable to determine qdiskd votes " "automatically\n"); else - logt_print(LOG_DEBUG, "Setting votes to %d\n", ret); + logt_print(LOG_DEBUG, "Setting autocalculated votes to %d\n", ret); return (ret); } @@ -1667,6 +1667,8 @@ get_dynamic_config_data(qd_ctx *ctx, int ccsfd) ctx->qc_flags &= ~RF_AUTO_VOTES; } + ctx->qc_auto_votes = auto_qdisk_votes(ccsfd); + snprintf(query, sizeof(query), "/cluster/quorumd/@votes"); if (ccs_get(ccsfd, query, &val) == 0) { ctx->qc_votes = atoi(val); @@ -1674,7 +1676,7 @@ get_dynamic_config_data(qd_ctx *ctx, int ccsfd) if (ctx->qc_votes < 0) ctx->qc_votes = 0; } else { - ctx->qc_votes = auto_qdisk_votes(ccsfd); + ctx->qc_votes = ctx->qc_auto_votes; if (ctx->qc_votes < 0) { if (ctx->qc_config) { logt_print(LOG_WARNING, "Unable to determine " @@ -1940,15 +1942,21 @@ get_config_data(qd_ctx *ctx, struct h_data *h, int maxh, int *cfh) *cfh = configure_heuristics(ccsfd, h, maxh, ctx->qc_interval * (ctx->qc_tko - 1)); - if (*cfh) { - if (ctx->qc_flags & RF_MASTER_WINS) { - logt_print(LOG_WARNING, "Master-wins mode disabled\n"); + if (ctx->qc_flags & RF_MASTER_WINS) { + if (*cfh) { + logt_print(LOG_WARNING, "Master-wins mode disabled " + "(not compatible with heuristics)\n"); + ctx->qc_flags &= ~RF_MASTER_WINS; + } + if (ctx->qc_auto_votes != 1) { + logt_print(LOG_WARNING, "Master-wins mode disabled " + "(not compatible with more than 2 nodes)\n"); ctx->qc_flags &= ~RF_MASTER_WINS; } } else { if (ctx->qc_flags & RF_AUTO_VOTES && - !(ctx->qc_flags & RF_MASTER_WINS) && - ctx->qc_votes == 1) { + !*cfh && + ctx->qc_auto_votes == 1) { /* Two node cluster, no heuristics, 1 vote for * quorum disk daemon. Safe to enable master-wins. * In fact, qdiskd without master-wins in this config

1 0

fence-agents: master - fence_ilo2 was not correctly cleaned by Makefile
by Marek Grác 26 Jul '12

26 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=cd73f094… Commit: cd73f094f7c163aca57a1dc9d5138fb468846835 Parent: 87266bcbf3f1b71664e65d82fb6de9f8814cb19e Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Thu Jul 26 14:35:44 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Thu Jul 26 14:35:44 2012 +0200 fence_ilo2 was not correctly cleaned by Makefile Related: rhbz#800650 --- fence/agents/ilo/Makefile.am | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/fence/agents/ilo/Makefile.am b/fence/agents/ilo/Makefile.am index 120727a..8df9a39 100644 --- a/fence/agents/ilo/Makefile.am +++ b/fence/agents/ilo/Makefile.am @@ -20,4 +20,4 @@ include $(top_srcdir)/make/fencebuild.mk include $(top_srcdir)/make/fenceman.mk clean-local: clean-man - rm -f $(TARGET) + rm -f $(TARGET) $(SYMTARGET)

1 0

fence-agents: master - Add fence agents specific for HP iLO2, iLO3, IMM and iDrac
by Marek Grác 26 Jul '12

26 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=87266bcb… Commit: 87266bcbf3f1b71664e65d82fb6de9f8814cb19e Parent: dff6b12ab73c8c38fc579ec312f398d6711cfeb8 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Thu Jul 26 14:27:14 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Thu Jul 26 14:27:14 2012 +0200 Add fence agents specific for HP iLO2, iLO3, IMM and iDrac These fence agents runs 'standard' fence agent (fence_ilo, fence_ipmilan) with correct timeout options, encryption, snmp_version, ... Originally they were supposed to be symbolic links but this approach is not recommended so they regular files. Resolves: rhbz#800650 --- fence/agents/ilo/Makefile.am | 10 ++++++++-- fence/agents/ilo/fence_ilo.py | 1 + fence/agents/ipmilan/Makefile.am | 11 +++++++++-- fence/agents/ipmilan/ipmilan.c | 10 ++++++++++ fence/agents/lib/fence2man.xsl | 4 ++++ fence/agents/lib/fencing.py.py | 3 +++ 6 files changed, 35 insertions(+), 4 deletions(-) diff --git a/fence/agents/ilo/Makefile.am b/fence/agents/ilo/Makefile.am index 7b44f59..120727a 100644 --- a/fence/agents/ilo/Makefile.am +++ b/fence/agents/ilo/Makefile.am @@ -2,13 +2,19 @@ MAINTAINERCLEANFILES = Makefile.in TARGET = fence_ilo +SYMTARGET = fence_ilo2 + SRC = $(TARGET).py EXTRA_DIST = $(SRC) -sbin_SCRIPTS = $(TARGET) +sbin_SCRIPTS = $(TARGET) $(SYMTARGET) + +man_MANS = $(TARGET).8 $(SYMTARGET).8 + +$(SYMTARGET): $(TARGET) + cp $^ $@ -man_MANS = $(TARGET).8 include $(top_srcdir)/make/fencebuild.mk include $(top_srcdir)/make/fenceman.mk diff --git a/fence/agents/ilo/fence_ilo.py b/fence/agents/ilo/fence_ilo.py index f70dc9c..d4e34df 100644 --- a/fence/agents/ilo/fence_ilo.py +++ b/fence/agents/ilo/fence_ilo.py @@ -82,6 +82,7 @@ The agent opens an SSL connection to the iLO card. Once the SSL \ connection is established, the agent is able to communicate with \ the iLO card through an XML stream." docs["vendorurl"] = "http://www.hp.com" + docs["symlink"] = [("fence_ilo2", "Fence agent for HP iLO2")] show_docs(options, docs) ## diff --git a/fence/agents/ipmilan/Makefile.am b/fence/agents/ipmilan/Makefile.am index 17ac8f2..c4fc636 100644 --- a/fence/agents/ipmilan/Makefile.am +++ b/fence/agents/ipmilan/Makefile.am @@ -2,15 +2,22 @@ MAINTAINERCLEANFILES = Makefile.in TARGET = fence_ipmilan +SYMTARGET = fence_ilo3 fence_imm fence_idrac + sbin_PROGRAMS = $(TARGET) +sbin_SCRIPTS = $(SYMTARGET) + noinst_HEADERS = expect.h fence_ipmilan_SOURCES = expect.c ipmilan.c -man_MANS = $(TARGET).8 +man_MANS = $(TARGET).8 $(SYMTARGET:%=%.8) + +$(SYMTARGET): $(TARGET) + cp $^ $@ include $(top_srcdir)/make/fencemanc.mk clean-local: clean-man - rm -f $(TARGET) + rm -f $(TARGET) $(SYMTARGET) diff --git a/fence/agents/ipmilan/ipmilan.c b/fence/agents/ipmilan/ipmilan.c index ef2b582..d1a0530 100644 --- a/fence/agents/ipmilan/ipmilan.c +++ b/fence/agents/ipmilan/ipmilan.c @@ -903,6 +903,9 @@ static void print_xml_metadata(char *pname) { printf("%s\n","<?xml version=\"1.0\" ?>"); printf("%s%s%s\n","<resource-agent name=\"",pname,"\" shortdesc=\"Fence agent for IPMI over LAN\">"); + printf("<symlink name=\"fence_ilo3\" shortdesc=\"Fence agent for HP iLO2\"/>\n"); + printf("<symlink name=\"fence_idrac\" shortdesc=\"Fence agent for Dell iDRAC\"/>\n"); + printf("<symlink name=\"fence_imm\" shortdesc=\"Fence agent for IBM Integrated Management Module\"/>\n"); printf("<longdesc>\n"); printf("fence_ipmilan is an I/O Fencing agent which can be used with " "machines controlled by IPMI. This agent calls support software " @@ -967,6 +970,7 @@ main(int argc, char **argv) int cipher=-1; int print_final_status=1; int translated_ret = -1; + char *filename; memset(ip, 0, sizeof(ip)); memset(authtype, 0, sizeof(authtype)); @@ -977,6 +981,12 @@ main(int argc, char **argv) memset(method, 0, sizeof(method)); memset(delay, 0, sizeof(delay)); + if (!strcmp(pname, "fence_ilo3")) { + lanplus = 1; + down_sleep = 4; + strncpy(method, "cycle\0", 6); + } + if (argc > 1) { /* Parse command line options if any were specified diff --git a/fence/agents/lib/fence2man.xsl b/fence/agents/lib/fence2man.xsl index 0fd71be..89baf7f 100644 --- a/fence/agents/lib/fence2man.xsl +++ b/fence/agents/lib/fence2man.xsl @@ -33,6 +33,10 @@ .TH FENCE_AGENT 8 2009-10-20 "<xsl:value-of select="@name"/> (Fence Agent)" .SH NAME <xsl:value-of select="@name" /> - <xsl:value-of select="@shortdesc" /> +<xsl:for-each select="symlink"> +.P +<xsl:value-of select="@name" /> - <xsl:value-of select="@shortdesc" /> (symlink) +</xsl:for-each> .SH DESCRIPTION .P <xsl:value-of select="longdesc"/> diff --git a/fence/agents/lib/fencing.py.py b/fence/agents/lib/fencing.py.py index a9c287a..cb8ec26 100644 --- a/fence/agents/lib/fencing.py.py +++ b/fence/agents/lib/fencing.py.py @@ -471,6 +471,9 @@ def metadata(avail_opt, options, docs): print "<?xml version=\"1.0\" ?>" print "<resource-agent name=\"" + os.path.basename(sys.argv[0]) + "\" shortdesc=\"" + docs["shortdesc"] + "\" >" + if "symlink" in docs: + for (symlink,desc) in docs["symlink"]: + print "<symlink name=\"" + symlink + "\" shortdesc=\"" + desc + "\"/>" print "<longdesc>" + docs["longdesc"] + "</longdesc>" if docs.has_key("vendorurl"): print "<vendor-url>" + docs["vendorurl"] + "</vendor-url>"

1 0

cluster: RHEL6 - cman init: allow dlm hash table sizes to be tunable at startup
by Fabio M. Di Nitto 26 Jul '12

26 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=982f2aa377fd1… Commit: 982f2aa377fd1c0471b714945c401fdfe86e9bc1 Parent: 88e4349b9b636d12b3c51f9c0f734b0259b36954 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Wed Jul 25 07:21:19 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Thu Jul 26 13:23:05 2012 +0200 cman init: allow dlm hash table sizes to be tunable at startup Resolves: rhbz#842370 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Acked-by: David Teigland <teigland(a)redhat.com> --- cman/init.d/cman.in | 28 ++++++++++++++++++++++++++++ cman/init.d/cman.init.defaults.in | 7 +++++++ 2 files changed, 35 insertions(+), 0 deletions(-) diff --git a/cman/init.d/cman.in b/cman/init.d/cman.in index 9a0d726..9de349d 100644 --- a/cman/init.d/cman.in +++ b/cman/init.d/cman.in @@ -110,6 +110,13 @@ fi # DLM_CONTROLD_OPTS -- allow extra options to be passed to dlm_controld daemon. [ -z "$DLM_CONTROLD_OPTS" ] && DLM_CONTROLD_OPTS="" +# DLM_LKBTBL_SIZE - DLM_RSBTBL_SIZE - DLM_DIRTBL_SIZE +# Allow tuning of DLM kernel hash table sizes. +# do NOT change unless instructed to do so. +[ -z "$DLM_LKBTBL_SIZE" ] && DLM_LKBTBL_SIZE="" +[ -z "$DLM_RSBTBL_SIZE" ] && DLM_RSBTBL_SIZE="" +[ -z "$DLM_DIRTBL_SIZE" ] && DLM_DIRTBL_SIZE="" + # FENCE_JOIN_TIMEOUT -- seconds to wait for fence domain join to # complete. If the join hasn't completed in this time, fence_tool join # exits with an error, and this script exits with an error. To wait @@ -706,6 +713,23 @@ leave_fence_domain() fi } +tune_dlm_hash_sizes() +{ + dlmdir="/sys/kernel/config/dlm/cluster" + + [ -n "$DLM_LKBTBL_SIZE" ] && [ -f $dlmdir/lkbtbl_size ] && \ + echo $DLM_LKBTBL_SIZE > $dlmdir/lkbtbl_size + + [ -n "$DLM_RSBTBL_SIZE" ] && [ -f $dlmdir/rsbtbl_size ] && \ + echo $DLM_RSBTBL_SIZE > $dlmdir/rsbtbl_size + + [ -n "$DLM_DIRTBL_SIZE" ] && [ -f $dlmdir/dirtbl_size ] && \ + echo $DLM_DIRTBL_SIZE > $dlmdir/dirtbl_size + + return 0 +} + + start() { currentaction="start" @@ -773,6 +797,10 @@ start() none \ "Starting dlm_controld" + runwrap tune_dlm_hash_sizes \ + none \ + "Tuning DLM kernel hash tables" + runwrap start_gfs_controld \ none \ "Starting gfs_controld" diff --git a/cman/init.d/cman.init.defaults.in b/cman/init.d/cman.init.defaults.in index 1b7913e..bbaa049 100644 --- a/cman/init.d/cman.init.defaults.in +++ b/cman/init.d/cman.init.defaults.in @@ -34,6 +34,13 @@ # DLM_CONTROLD_OPTS -- allow extra options to be passed to dlm_controld daemon. #DLM_CONTROLD_OPTS="" +# DLM_LKBTBL_SIZE - DLM_RSBTBL_SIZE - DLM_DIRTBL_SIZE +# Allow tuning of DLM kernel hash table sizes. +# do NOT change unless instructed to do so. +#DLM_LKBTBL_SIZE="" +#DLM_RSBTBL_SIZE="" +#DLM_DIRTBL_SIZE="" + # FENCE_JOIN_TIMEOUT -- seconds to wait for fence domain join to # complete. If the join hasn't completed in this time, fence_tool join # exits with an error, and this script exits with an error. To wait

1 0

cluster: RHEL59 - fence_xvm: Fix incorrect success return codes
by Ryan McCabe 25 Jul '12

25 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=0a588f475afdb… Commit: 0a588f475afdb78660b424587f2c6460d4f1be82 Parent: 9eccf2611b6bfd92e4b15196b4744f24991a7bb8 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Wed Jul 25 10:52:33 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Wed Jul 25 10:52:33 2012 -0400 fence_xvm: Fix incorrect success return codes This patch fixes two places where fence_xvm will return success for fencing operations when fencing has failed due to the challenge/response failing. Resolves: rhbz#843083 Signed-off-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- fence/agents/xvm/fence_xvm.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/fence/agents/xvm/fence_xvm.c b/fence/agents/xvm/fence_xvm.c index d1ffb12..eb006fd 100644 --- a/fence/agents/xvm/fence_xvm.c +++ b/fence/agents/xvm/fence_xvm.c @@ -99,14 +99,14 @@ tcp_exchange(int fd, fence_auth_type_t auth, void *key, if (tcp_challenge(fd, auth, key, key_len, timeout) <= 0) { /* Challenge failed */ printf("Invalid response to challenge\n"); - return 0; + return 1; } /* Now they'll send us one, so we need to respond here */ dbg_printf(3, "Responding to TCP challenge\n"); if (tcp_response(fd, auth, key, key_len, timeout) <= 0) { printf("Invalid response to challenge\n"); - return 0; + return 1; } dbg_printf(2, "TCP Exchange + Authentication done... \n");

1 0

cluster: STABLE32 - cman init: allow dlm hash table sizes to be tunable at startup
by Fabio M. Di Nitto 25 Jul '12

25 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=c65a13a82189f… Commit: c65a13a82189f38bd80d5c52924d09a7d98daa37 Parent: 224a841585ea300718f9e82d9fadd0f163233e8f Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Wed Jul 25 07:21:19 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Wed Jul 25 07:21:19 2012 +0200 cman init: allow dlm hash table sizes to be tunable at startup Resolves: rhbz#842370 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/init.d/cman.in | 28 ++++++++++++++++++++++++++++ cman/init.d/cman.init.defaults.in | 7 +++++++ 2 files changed, 35 insertions(+), 0 deletions(-) diff --git a/cman/init.d/cman.in b/cman/init.d/cman.in index c6e038e..64d12de 100644 --- a/cman/init.d/cman.in +++ b/cman/init.d/cman.in @@ -119,6 +119,13 @@ fi # DLM_CONTROLD_OPTS -- allow extra options to be passed to dlm_controld daemon. [ -z "$DLM_CONTROLD_OPTS" ] && DLM_CONTROLD_OPTS="" +# DLM_LKBTBL_SIZE - DLM_RSBTBL_SIZE - DLM_DIRTBL_SIZE +# Allow tuning of DLM kernel hash table sizes. +# do NOT change unless instructed to do so. +[ -z "$DLM_LKBTBL_SIZE" ] && DLM_LKBTBL_SIZE="" +[ -z "$DLM_RSBTBL_SIZE" ] && DLM_RSBTBL_SIZE="" +[ -z "$DLM_DIRTBL_SIZE" ] && DLM_DIRTBL_SIZE="" + # FENCE_JOIN_TIMEOUT -- seconds to wait for fence domain join to # complete. If the join hasn't completed in this time, fence_tool join # exits with an error, and this script exits with an error. To wait @@ -755,6 +762,23 @@ leave_fence_domain() fi } +tune_dlm_hash_sizes() +{ + dlmdir="/sys/kernel/config/dlm/cluster" + + [ -n "$DLM_LKBTBL_SIZE" ] && [ -f $dlmdir/lkbtbl_size ] && \ + echo $DLM_LKBTBL_SIZE > $dlmdir/lkbtbl_size + + [ -n "$DLM_RSBTBL_SIZE" ] && [ -f $dlmdir/rsbtbl_size ] && \ + echo $DLM_RSBTBL_SIZE > $dlmdir/rsbtbl_size + + [ -n "$DLM_DIRTBL_SIZE" ] && [ -f $dlmdir/dirtbl_size ] && \ + echo $DLM_DIRTBL_SIZE > $dlmdir/dirtbl_size + + return 0 +} + + start() { currentaction="start" @@ -822,6 +846,10 @@ start() dlm_controld_enabled \ "Starting dlm_controld" + runwrap tune_dlm_hash_sizes \ + none \ + "Tuning DLM kernel hash tables" + gfs_controld_enabled && cd @INITDDIR@ && ./gfs2-cluster start runwrap start_ocfs2_controld \ diff --git a/cman/init.d/cman.init.defaults.in b/cman/init.d/cman.init.defaults.in index 5be97ff..03b2474 100644 --- a/cman/init.d/cman.init.defaults.in +++ b/cman/init.d/cman.init.defaults.in @@ -42,6 +42,13 @@ # DLM_CONTROLD_OPTS -- allow extra options to be passed to dlm_controld daemon. #DLM_CONTROLD_OPTS="" +# DLM_LKBTBL_SIZE - DLM_RSBTBL_SIZE - DLM_DIRTBL_SIZE +# Allow tuning of DLM kernel hash table sizes. +# do NOT change unless instructed to do so. +#DLM_LKBTBL_SIZE="" +#DLM_RSBTBL_SIZE="" +#DLM_DIRTBL_SIZE="" + # FENCE_JOIN_TIMEOUT -- seconds to wait for fence domain join to # complete. If the join hasn't completed in this time, fence_tool join # exits with an error, and this script exits with an error. To wait

1 0

cluster: RHEL59 - rgmanager: Add IP resource agent "prefer_interface" parameter
by Ryan McCabe 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=9eccf2611b6bf… Commit: 9eccf2611b6bfd92e4b15196b4744f24991a7bb8 Parent: 332db7754d6210991c34be056789aa1b33b26a0a Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Tue Jul 24 13:34:35 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Tue Jul 24 18:13:09 2012 -0400 rgmanager: Add IP resource agent "prefer_interface" parameter This patch adds a "prefer_interface" parameter for IP resources. The interface must already be configured and active. This parameter should be used only when at least two active interfaces have IP addresses on the same subnet and it's necessary to specify which particular interface should be used. Acked-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/resources/ip.sh | 17 +++++++++++++++++ 1 files changed, 17 insertions(+), 0 deletions(-) diff --git a/rgmanager/src/resources/ip.sh b/rgmanager/src/resources/ip.sh index 38d1ab9..bbd85f3 100755 --- a/rgmanager/src/resources/ip.sh +++ b/rgmanager/src/resources/ip.sh @@ -132,6 +132,15 @@ meta_data() <content type="boolean"/> </parameter> + <parameter name="prefer_interface"> + <longdesc lang="en"> + The network interface to which the IP address should be added. The interface must already be configured and active. This parameter should be used only when at least two active interfaces have IP addresses on the same subnet and it is desired to have the IP address added to a particular interface. + </longdesc> + <shortdesc lang="en"> + Network interface + </shortdesc> + <content type="string"/> + </parameter> </parameters> <actions> @@ -587,6 +596,10 @@ ipv6() fi if [ "$1" = "add" ]; then + if [ -n "$OCF_RESKEY_prefer_interface" ] && \ + [ "$OCF_RESKEY_prefer_interface" != $dev ]; then + continue + fi ipv6_same_subnet $ifaddr_exp/$maskbits $addr_exp if [ $? -ne 0 ]; then continue @@ -670,6 +683,10 @@ ipv4() fi if [ "$1" = "add" ]; then + if [ -n "$OCF_RESKEY_prefer_interface" ] && \ + [ "$OCF_RESKEY_prefer_interface" != $dev ]; then + continue + fi ipv4_same_subnet $ifaddr/$maskbits $addr if [ $? -ne 0 ]; then continue

1 0

cluster: RHEL6 - cman init: clarify better comments for FENCE_JOIN=no
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=88e4349b9b636… Commit: 88e4349b9b636d12b3c51f9c0f734b0259b36954 Parent: 0eecc716d53b4fae15ad131c5d10c628348e0012 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 11:24:39 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 11:31:39 2012 +0200 cman init: clarify better comments for FENCE_JOIN=no (input from GSS) Related: rhbz#821016 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Christine Caulfield <ccaulfie(a)redhat.com> --- cman/init.d/cman.in | 7 ++++--- cman/init.d/cman.init.defaults.in | 7 ++++--- 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/cman/init.d/cman.in b/cman/init.d/cman.in index f39a9aa..9a0d726 100644 --- a/cman/init.d/cman.in +++ b/cman/init.d/cman.in @@ -126,9 +126,10 @@ fi # set to "yes", then the script will attempt to join the fence domain. # If FENCE_JOIN is set to any other value, the default behavior is # to join the fence domain (equivalent to "yes"). -# When setting FENCE_JOIN to "no", it is important to check -# DLM_CONTROLD_OPTS to reflect expected behavior regarding fencing -# and quorum. +# When setting FENCE_JOIN to "no", it is important to also set +# DLM_CONTROLD_OPTS="-f0" (at least) for correct operation. +# Please note that clusters without fencing are not +# supported by Red Hat except for MRG installations. [ -z "$FENCE_JOIN" ] && FENCE_JOIN="yes" # FENCED_OPTS -- allow extra options to be passed to fence daemon. diff --git a/cman/init.d/cman.init.defaults.in b/cman/init.d/cman.init.defaults.in index 6e7f47f..1b7913e 100644 --- a/cman/init.d/cman.init.defaults.in +++ b/cman/init.d/cman.init.defaults.in @@ -50,9 +50,10 @@ # set to "yes", then the script will attempt to join the fence domain. # If FENCE_JOIN is set to any other value, the default behavior is # to join the fence domain (equivalent to "yes"). -# When setting FENCE_JOIN to "no", it is important to check -# DLM_CONTROLD_OPTS to reflect expected behavior regarding fencing -# and quorum. +# When setting FENCE_JOIN to "no", it is important to also set +# DLM_CONTROLD_OPTS="-f0" (at least) for correct operation. +# Please note that clusters without fencing are not +# supported by Red Hat except for MRG installations. #FENCE_JOIN="yes" # FENCED_OPTS -- allow extra options to be passed to fence daemon.

1 0

cluster: RHEL6 - cman init: add extra documentation for FENCE_JOIN=""
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=0eecc716d53b4… Commit: 0eecc716d53b4fae15ad131c5d10c628348e0012 Parent: 69d2920745d3ac11a44b5d9ce3b07e8eb1d07b25 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue May 15 16:37:13 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 11:29:59 2012 +0200 cman init: add extra documentation for FENCE_JOIN="" Related: rhbz#821016 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Christine Caulfield <ccaulfie(a)redhat.com> --- cman/init.d/cman.in | 3 +++ cman/init.d/cman.init.defaults.in | 3 +++ 2 files changed, 6 insertions(+), 0 deletions(-) diff --git a/cman/init.d/cman.in b/cman/init.d/cman.in index f345bcc..f39a9aa 100644 --- a/cman/init.d/cman.in +++ b/cman/init.d/cman.in @@ -126,6 +126,9 @@ fi # set to "yes", then the script will attempt to join the fence domain. # If FENCE_JOIN is set to any other value, the default behavior is # to join the fence domain (equivalent to "yes"). +# When setting FENCE_JOIN to "no", it is important to check +# DLM_CONTROLD_OPTS to reflect expected behavior regarding fencing +# and quorum. [ -z "$FENCE_JOIN" ] && FENCE_JOIN="yes" # FENCED_OPTS -- allow extra options to be passed to fence daemon. diff --git a/cman/init.d/cman.init.defaults.in b/cman/init.d/cman.init.defaults.in index d6d010c..6e7f47f 100644 --- a/cman/init.d/cman.init.defaults.in +++ b/cman/init.d/cman.init.defaults.in @@ -50,6 +50,9 @@ # set to "yes", then the script will attempt to join the fence domain. # If FENCE_JOIN is set to any other value, the default behavior is # to join the fence domain (equivalent to "yes"). +# When setting FENCE_JOIN to "no", it is important to check +# DLM_CONTROLD_OPTS to reflect expected behavior regarding fencing +# and quorum. #FENCE_JOIN="yes" # FENCED_OPTS -- allow extra options to be passed to fence daemon.

1 0

cluster: RHEL6 - cman init: allow sysconfig/cman to pass options to dlm_controld
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=69d2920745d3a… Commit: 69d2920745d3ac11a44b5d9ce3b07e8eb1d07b25 Parent: 293c73fc93fb82cc427c37df0c088ff39cf25f4c Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue May 15 13:53:28 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 11:28:58 2012 +0200 cman init: allow sysconfig/cman to pass options to dlm_controld DLM_CONTROLD_OPTS="" can now be used to pass startup options to the daemon. Resolves: rhbz#821016 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Christine Caulfield <ccaulfie(a)redhat.com> --- cman/init.d/cman.in | 5 ++++- cman/init.d/cman.init.defaults.in | 3 +++ 2 files changed, 7 insertions(+), 1 deletions(-) diff --git a/cman/init.d/cman.in b/cman/init.d/cman.in index 1268b59..f345bcc 100644 --- a/cman/init.d/cman.in +++ b/cman/init.d/cman.in @@ -107,6 +107,9 @@ fi # no (default) | cman will not start sshd [ -z "$CMAN_SSHD_START" ] && CMAN_SSHD_START=no +# DLM_CONTROLD_OPTS -- allow extra options to be passed to dlm_controld daemon. +[ -z "$DLM_CONTROLD_OPTS" ] && DLM_CONTROLD_OPTS="" + # FENCE_JOIN_TIMEOUT -- seconds to wait for fence domain join to # complete. If the join hasn't completed in this time, fence_tool join # exits with an error, and this script exits with an error. To wait @@ -615,7 +618,7 @@ stop_fenced() start_dlm_controld() { - start_daemon dlm_controld || return 1 + start_daemon dlm_controld "$DLM_CONTROLD_OPTS" || return 1 if [ "$INITLOGLEVEL" = "full" ]; then ok diff --git a/cman/init.d/cman.init.defaults.in b/cman/init.d/cman.init.defaults.in index 05c3bc7..d6d010c 100644 --- a/cman/init.d/cman.init.defaults.in +++ b/cman/init.d/cman.init.defaults.in @@ -31,6 +31,9 @@ # no (default) | cman will not start sshd #CMAN_SSHD_START=no +# DLM_CONTROLD_OPTS -- allow extra options to be passed to dlm_controld daemon. +#DLM_CONTROLD_OPTS="" + # FENCE_JOIN_TIMEOUT -- seconds to wait for fence domain join to # complete. If the join hasn't completed in this time, fence_tool join # exits with an error, and this script exits with an error. To wait

1 0

cluster: STABLE32 - cman init: clarify better comments for FENCE_JOIN=no
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=224a841585ea3… Commit: 224a841585ea300718f9e82d9fadd0f163233e8f Parent: e8af462b7531f87b5cb20f7204eeb4b520591da9 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 11:24:39 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 11:24:39 2012 +0200 cman init: clarify better comments for FENCE_JOIN=no (input from GSS) Related: rhbz#821016 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/init.d/cman.in | 7 ++++--- cman/init.d/cman.init.defaults.in | 7 ++++--- 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/cman/init.d/cman.in b/cman/init.d/cman.in index 95323b4..c6e038e 100644 --- a/cman/init.d/cman.in +++ b/cman/init.d/cman.in @@ -135,9 +135,10 @@ fi # set to "yes", then the script will attempt to join the fence domain. # If FENCE_JOIN is set to any other value, the default behavior is # to join the fence domain (equivalent to "yes"). -# When setting FENCE_JOIN to "no", it is important to check -# DLM_CONTROLD_OPTS to reflect expected behavior regarding fencing -# and quorum. +# When setting FENCE_JOIN to "no", it is important to also set +# DLM_CONTROLD_OPTS="-f0" (at least) for correct operation. +# Please note that clusters without fencing are not +# supported by Red Hat except for MRG installations. [ -z "$FENCE_JOIN" ] && FENCE_JOIN="yes" # FENCED_OPTS -- allow extra options to be passed to fence daemon. diff --git a/cman/init.d/cman.init.defaults.in b/cman/init.d/cman.init.defaults.in index b981bab..5be97ff 100644 --- a/cman/init.d/cman.init.defaults.in +++ b/cman/init.d/cman.init.defaults.in @@ -58,9 +58,10 @@ # set to "yes", then the script will attempt to join the fence domain. # If FENCE_JOIN is set to any other value, the default behavior is # to join the fence domain (equivalent to "yes"). -# When setting FENCE_JOIN to "no", it is important to check -# DLM_CONTROLD_OPTS to reflect expected behavior regarding fencing -# and quorum. +# When setting FENCE_JOIN to "no", it is important to also set +# DLM_CONTROLD_OPTS="-f0" (at least) for correct operation. +# Please note that clusters without fencing are not +# supported by Red Hat except for MRG installations. #FENCE_JOIN="yes" # FENCED_OPTS -- allow extra options to be passed to fence daemon.

1 0

cluster: RHEL6 - cman: fix a couple of comments
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=293c73fc93fb8… Commit: 293c73fc93fb82cc427c37df0c088ff39cf25f4c Parent: fa9267535f7bbb49dcc766f87f35502dfa623847 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Thu Jul 12 13:00:27 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 11:01:01 2012 +0200 cman: fix a couple of comments Related: rhbz#839241 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Christine Caulfield <ccaulfie(a)redhat.com> --- cman/daemon/cman-preconfig.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index 321d78f..99ee80d 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -1538,7 +1538,7 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const objdb->object_destroy(cluster_parent_handle); /* - * create cluster.cman in the new config if it doesn't exists + * create cluster.cman in the new config if it doesn't exist */ objdb->object_find_create(cluster_parent_handle_new, "cman", strlen("cman"), &find_handle); if (objdb->object_find_next(find_handle, &object_handle)) { @@ -1548,7 +1548,7 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const objdb->object_find_destroy(find_handle); /* - * readd cluster_id/two_node/nodename + * write cluster_id/two_node/nodename */ snprintf(str, sizeof(str) - 1, "%d", cluster_id); objdb->object_key_create_typed(object_handle, "cluster_id",

1 0

cluster: RHEL6 - cman: fix data copy and memory leak when reloading config
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=fa9267535f7bb… Commit: fa9267535f7bbb49dcc766f87f35502dfa623847 Parent: bdbed1f76496e2f3cfc86535314cbeffd4d20257 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Wed Jul 11 11:44:49 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 11:00:04 2012 +0200 cman: fix data copy and memory leak when reloading config cman.cluster_id,nodename,two_node information were not copied from the old to the new config at reload time. This triggers a problem when <cman is set in cluster.conf and we effectively drop information from objdb (suboptimal). Also fix a possible memory leak when we have reload issues. Resolves: rhbz#839241 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Christine Caulfield <ccaulfie(a)redhat.com> --- cman/daemon/cman-preconfig.c | 31 +++++++++++++++++++++++++++++-- 1 files changed, 29 insertions(+), 2 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index c42052e..321d78f 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -1483,6 +1483,7 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const hdb_handle_t cluster_parent_handle_new; unsigned int config_version = 0, config_version_new = 0; char *config_value = NULL; + char str[255]; /* don't reload if we've been told to run configless */ if (getenv("CMAN_NOCONFIG")) { @@ -1494,16 +1495,16 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const /* find both /cluster entries */ objdb->object_find_create(OBJECT_PARENT_HANDLE, "cluster", strlen("cluster"), &find_handle); objdb->object_find_next(find_handle, &cluster_parent_handle); + objdb->object_find_next(find_handle, &cluster_parent_handle_new); + objdb->object_find_destroy(find_handle); if (!cluster_parent_handle) { sprintf (error_reason, "%s", "Cannot find old /cluster/ key in configuration\n"); goto err; } - objdb->object_find_next(find_handle, &cluster_parent_handle_new); if (!cluster_parent_handle_new) { sprintf (error_reason, "%s", "Cannot find new /cluster/ key in configuration\n"); goto err; } - objdb->object_find_destroy(find_handle); if (!objdb->object_key_get(cluster_parent_handle, "config_version", strlen("config_version"), (void *)&config_value, NULL)) { if (config_value) { @@ -1536,6 +1537,32 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const /* destroy the old one */ objdb->object_destroy(cluster_parent_handle); + /* + * create cluster.cman in the new config if it doesn't exists + */ + objdb->object_find_create(cluster_parent_handle_new, "cman", strlen("cman"), &find_handle); + if (objdb->object_find_next(find_handle, &object_handle)) { + objdb->object_create(cluster_parent_handle_new, &object_handle, + "cman", strlen("cman")); + } + objdb->object_find_destroy(find_handle); + + /* + * readd cluster_id/two_node/nodename + */ + snprintf(str, sizeof(str) - 1, "%d", cluster_id); + objdb->object_key_create_typed(object_handle, "cluster_id", + str, strlen(str) + 1, OBJDB_VALUETYPE_STRING); + + if (two_node) { + snprintf(str, sizeof(str) - 1, "%d", 1); + objdb->object_key_create_typed(object_handle, "two_node", + str, strlen(str) + 1, OBJDB_VALUETYPE_STRING); + } + + objdb->object_key_create_typed(object_handle, "nodename", + nodename, strlen(nodename)+1, OBJDB_VALUETYPE_STRING); + /* update the reference to the new config */ cluster_parent_handle = cluster_parent_handle_new;

1 0

cluster: RHEL6 - cman-preconfig: allow host aliases as valid cluster nodenames
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=bdbed1f76496e… Commit: bdbed1f76496e2f3cfc86535314cbeffd4d20257 Parent: af86e47803e3626641c1671ac89437d4a82066a7 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Wed Jun 27 11:46:01 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 10:47:40 2012 +0200 cman-preconfig: allow host aliases as valid cluster nodenames Resolves: rhbz#786118 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Christine Caulfield <ccaulfie(a)redhat.com> --- cman/daemon/cman-preconfig.c | 91 +++++++++++++++++++++++++++++++++++------- 1 files changed, 76 insertions(+), 15 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index c8f69e5..c42052e 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -462,7 +462,7 @@ static int verify_nodename(struct objdb_iface_ver0 *objdb, char *node) struct sockaddr *sa; hdb_handle_t nodes_handle; hdb_handle_t find_handle = 0; - int error; + int found = 0; /* nodename is either from commandline or from uname */ if (nodelist_byname(objdb, cluster_parent_handle, node)) @@ -508,12 +508,11 @@ static int verify_nodename(struct objdb_iface_ver0 *objdb, char *node) } while (nodes_handle); objdb->object_find_destroy(find_handle); - - /* The cluster.conf names may not be related to uname at all, - they may match a hostname on some network interface. - NOTE: This is IPv4 only */ - error = getifaddrs(&ifa_list); - if (error) + /* + * The cluster.conf names may not be related to uname at all, + * they may match a hostname on some network interface. + */ + if (getifaddrs(&ifa_list)) return -1; for (ifa = ifa_list; ifa; ifa = ifa->ifa_next) { @@ -532,12 +531,13 @@ static int verify_nodename(struct objdb_iface_ver0 *objdb, char *node) if (sa->sa_family == AF_INET6) salen = sizeof(struct sockaddr_in6); - error = getnameinfo(sa, salen, nodename2, - sizeof(nodename2), NULL, 0, 0); - if (!error) { + if (getnameinfo(sa, salen, + nodename2, sizeof(nodename2), + NULL, 0, 0) == 0) { if (nodelist_byname(objdb, cluster_parent_handle, nodename2)) { strcpy(node, nodename2); + found = 1; goto out; } @@ -548,27 +548,88 @@ static int verify_nodename(struct objdb_iface_ver0 *objdb, char *node) if (nodelist_byname(objdb, cluster_parent_handle, nodename2)) { strcpy(node, nodename2); + found = 1; goto out; } } } /* See if it's the IP address that's in cluster.conf */ - error = getnameinfo(sa, sizeof(*sa), nodename2, - sizeof(nodename2), NULL, 0, NI_NUMERICHOST); - if (error) + if (getnameinfo(sa, sizeof(*sa), + nodename2, sizeof(nodename2), + NULL, 0, NI_NUMERICHOST)) continue; if (nodelist_byname(objdb, cluster_parent_handle, nodename2)) { strcpy(node, nodename2); + found = 1; goto out; } } - error = -1; out: + if (found) { + freeifaddrs(ifa_list); + return 0; + } + + /* + * This section covers the usecase where the nodename specified in cluster.conf + * is an alias specified in /etc/hosts. For example: + * <ipaddr> hostname alias1 alias2 + * and <clusternode name="alias2"> + * the above calls use uname and getnameinfo does not return aliases. + * here we take the name specified in cluster.conf, resolve it to an address + * and then compare against all known local ip addresses. + * if we have a match, we found our nodename. In theory this chunk of code + * could replace all the checks above, but let's avoid any possible regressions + * and use it as last. + */ + + nodes_handle = nodeslist_init(objdb, cluster_parent_handle, &find_handle); + while (nodes_handle) { + char *dbnodename = NULL; + struct addrinfo hints; + struct addrinfo *result = NULL, *rp = NULL; + + if (objdb_get_string(objdb, nodes_handle, "name", &dbnodename)) { + goto next; + } + + memset(&hints, 0, sizeof(struct addrinfo)); + hints.ai_family = AF_UNSPEC; + hints.ai_socktype = SOCK_DGRAM; + hints.ai_flags = 0; + hints.ai_protocol = IPPROTO_UDP; + + if (getaddrinfo(dbnodename, NULL, &hints, &result)) + goto next; + + for (rp = result; rp != NULL; rp = rp->ai_next) { + for (ifa = ifa_list; ifa; ifa = ifa->ifa_next) { + if (ipaddr_equal((struct sockaddr_storage *)rp->ai_addr, + (struct sockaddr_storage *)ifa->ifa_addr)) { + freeaddrinfo(result); + strncpy(node, dbnodename, sizeof(nodename) - 1); + found = 1; + goto out2; + } + } + } + + freeaddrinfo(result); + next: + nodes_handle = nodeslist_next(objdb, find_handle); + } + out2: + objdb->object_find_destroy(find_handle); freeifaddrs(ifa_list); - return error; + + if (found) { + return 0; + } + + return -1; } /* Get any environment variable overrides */

1 0

cluster: RHEL6 - config: Fix typo in schema
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=af86e47803e36… Commit: af86e47803e3626641c1671ac89437d4a82066a7 Parent: 5674e99cd0b141c34630ec411e0d682727a30dc9 Author: Lon Hohberger <lhh(a)redhat.com> AuthorDate: Mon Jan 30 16:35:52 2012 -0500 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 10:39:23 2012 +0200 config: Fix typo in schema Resolves: rhbz#785866 Signed-off-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- config/tools/xml/cluster.rng.in.head | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/config/tools/xml/cluster.rng.in.head b/config/tools/xml/cluster.rng.in.head index 1bdb836..4e3d901 100644 --- a/config/tools/xml/cluster.rng.in.head +++ b/config/tools/xml/cluster.rng.in.head @@ -20,7 +20,7 @@ cluster - rm - clvmd -Element defnitions: +Element definitions: - Resource - Fence

1 0

cluster: RHEL6 - qdiskd: allow master to failover quickly when using master_wins
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=5674e99cd0b14… Commit: 5674e99cd0b141c34630ec411e0d682727a30dc9 Parent: e01a564837cc5f5310a046da3c71f995c301e649 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 10:27:57 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 10:34:25 2012 +0200 qdiskd: allow master to failover quickly when using master_wins in case of master_wins and we are shutting down the master qdiskd, there is a small window in which the other node is not quorate because qdiskd has not become master yet. this patch allows the master qdiskd to communicate to the other nodes that it is going away and gives enough time to elect a new master before dieing. the process itself is safe and the worst case scenario the cluster will behave as-is now (temporary loss of quorum), otherwise a fast switch will take place. Resolves: rhbz#814807 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Lon Hohberger <lhh(a)redhat.com> --- cman/qdisk/disk.h | 5 +++- cman/qdisk/main.c | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 75 insertions(+), 2 deletions(-) diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index fd80fa6..6bed41d 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -24,9 +24,12 @@ typedef enum { S_INIT = 0x2, // Initializing. Hold your fire. /* vvv Fencing will kill a node */ S_RUN = 0x5, // I think I'm running. - S_MASTER= 0x6 // I know I'm running, and have advertised to + S_MASTER= 0x6, // I know I'm running, and have advertised to // CMAN the availability of the disk vote for my // partition. + S_EXIT = 0x7 // trigger master re-election before exit + // status is set only by master in master-win | auto-masterwin + // and next status _must_ be S_NONE } disk_node_state_t; diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index fc714f6..594e9e8 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -194,7 +194,8 @@ read_node_blocks(qd_ctx *ctx, node_info_t *ni, int max) continue; /* Unchanged timestamp: miss */ - if (sb->ps_timestamp == ni[x].ni_last_seen) { + if ((sb->ps_timestamp == ni[x].ni_last_seen) && + (ni[x].ni_state != S_EXIT)) { /* XXX check for average + allow grace */ ni[x].ni_misses++; if (ni[x].ni_misses > 1) { @@ -229,6 +230,22 @@ check_transitions(qd_ctx *ctx, node_info_t *ni, int max, memb_mask_t mask) for (x = 0; x < max; x++) { /* + Case 0: check if master node is about to leave + */ + if (ni[x].ni_state == S_EXIT) { + logt_print(LOG_NOTICE, "Node %d is about to leave\n", ni[x].ni_status.ps_nodeid); + ni[x].ni_evil_incarnation = 0; + ni[x].ni_incarnation = 0; + ni[x].ni_seen = 0; + ni[x].ni_misses = 0; + ni[x].ni_state = S_NONE; + if (mask) + clear_bit(mask, (ni[x].ni_status.ps_nodeid-1), + sizeof(memb_mask_t)); + continue; + } + + /* Case 1: check to see if the node is still up according to our internal state, but has been evicted by the master or cleanly shut down @@ -1259,6 +1276,50 @@ quorum_loop(qd_ctx *ctx, node_info_t *ni, int max) /** + Tell the other nodes to elect a new master != me. + */ +static int +quorum_reelect_master(qd_ctx *ctx, node_info_t *ni, int max) +{ + if (qd_write_status(ctx, ctx->qc_my_id, S_EXIT, + NULL, NULL, NULL) != 0) { + logt_print(LOG_WARNING, + "Error writing to quorum disk during reelect_master\n"); + } + + while (1) { + int master, x; + int found = 0; + int low_id, count; + + read_node_blocks(ctx, ni, max); + + for (x = 0; x < max; x++) { + if (ni[x].ni_state >= S_RUN) { + found = 1; + } + } + + if (!found) { + logt_print(LOG_DEBUG, "No other nodes are active. Exiting\n"); + break; + } + + master = master_exists(ctx, ni, max, &low_id, &count); + if (master) { + logt_print(LOG_DEBUG, "New master elected: %d\n", master); + break; + } + /* + * give time for message to be read + */ + sleep(1); + } + + return 0; +} + +/** Tell the other nodes we're done (safely!). */ static int @@ -2151,6 +2212,15 @@ main(int argc, char **argv) io_nanny_start(ch_user, ctx.qc_tko * ctx.qc_interval); if (quorum_loop(&ctx, ni, MAX_NODES_DISK) == 0) { + /* + * if we are master and we are in master-win mode, + * request other qdiskd to elect a new one + */ + if ((ctx.qc_status == S_MASTER) && + ((ctx.qc_flags & RF_MASTER_WINS) || + (ctx.qc_flags & RF_AUTO_MASTER_WINS))) { + quorum_reelect_master(&ctx, ni, MAX_NODES_DISK); + } /* Only clean up if we're exiting w/o error) */ logt_print(LOG_NOTICE, "Unregistering quorum device.\n"); cman_unregister_quorum_device(ctx.qc_cman_admin);

1 0

cluster: STABLE32 - qdiskd: allow master to failover quickly when using master_wins
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=e8af462b7531f… Commit: e8af462b7531f87b5cb20f7204eeb4b520591da9 Parent: 03e2215bd277fd79b8a6ee70a49de711e0f343ad Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 10:27:57 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 10:27:57 2012 +0200 qdiskd: allow master to failover quickly when using master_wins in case of master_wins and we are shutting down the master qdiskd, there is a small window in which the other node is not quorate because qdiskd has not become master yet. this patch allows the master qdiskd to communicate to the other nodes that it is going away and gives enough time to elect a new master before dieing. the process itself is safe and the worst case scenario the cluster will behave as-is now (temporary loss of quorum), otherwise a fast switch will take place. Resolves: rhbz#814807 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Lon Hohberger <lhh(a)redhat.com> --- cman/qdisk/disk.h | 5 +++- cman/qdisk/main.c | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 75 insertions(+), 2 deletions(-) diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index fd80fa6..6bed41d 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -24,9 +24,12 @@ typedef enum { S_INIT = 0x2, // Initializing. Hold your fire. /* vvv Fencing will kill a node */ S_RUN = 0x5, // I think I'm running. - S_MASTER= 0x6 // I know I'm running, and have advertised to + S_MASTER= 0x6, // I know I'm running, and have advertised to // CMAN the availability of the disk vote for my // partition. + S_EXIT = 0x7 // trigger master re-election before exit + // status is set only by master in master-win | auto-masterwin + // and next status _must_ be S_NONE } disk_node_state_t; diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index 32677a2..16c26e4 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -196,7 +196,8 @@ read_node_blocks(qd_ctx *ctx, node_info_t *ni, int max) continue; /* Unchanged timestamp: miss */ - if (sb->ps_timestamp == ni[x].ni_last_seen) { + if ((sb->ps_timestamp == ni[x].ni_last_seen) && + (ni[x].ni_state != S_EXIT)) { /* XXX check for average + allow grace */ ni[x].ni_misses++; if (ni[x].ni_misses > 1) { @@ -231,6 +232,22 @@ check_transitions(qd_ctx *ctx, node_info_t *ni, int max, memb_mask_t mask) for (x = 0; x < max; x++) { /* + Case 0: check if master node is about to leave + */ + if (ni[x].ni_state == S_EXIT) { + logt_print(LOG_NOTICE, "Node %d is about to leave\n", ni[x].ni_status.ps_nodeid); + ni[x].ni_evil_incarnation = 0; + ni[x].ni_incarnation = 0; + ni[x].ni_seen = 0; + ni[x].ni_misses = 0; + ni[x].ni_state = S_NONE; + if (mask) + clear_bit(mask, (ni[x].ni_status.ps_nodeid-1), + sizeof(memb_mask_t)); + continue; + } + + /* Case 1: check to see if the node is still up according to our internal state, but has been evicted by the master or cleanly shut down @@ -1269,6 +1286,50 @@ quorum_loop(qd_ctx *ctx, node_info_t *ni, int max) /** + Tell the other nodes to elect a new master != me. + */ +static int +quorum_reelect_master(qd_ctx *ctx, node_info_t *ni, int max) +{ + if (qd_write_status(ctx, ctx->qc_my_id, S_EXIT, + NULL, NULL, NULL) != 0) { + logt_print(LOG_WARNING, + "Error writing to quorum disk during reelect_master\n"); + } + + while (1) { + int master, x; + int found = 0; + int low_id, count; + + read_node_blocks(ctx, ni, max); + + for (x = 0; x < max; x++) { + if (ni[x].ni_state >= S_RUN) { + found = 1; + } + } + + if (!found) { + logt_print(LOG_DEBUG, "No other nodes are active. Exiting\n"); + break; + } + + master = master_exists(ctx, ni, max, &low_id, &count); + if (master) { + logt_print(LOG_DEBUG, "New master elected: %d\n", master); + break; + } + /* + * give time for message to be read + */ + sleep(1); + } + + return 0; +} + +/** Tell the other nodes we're done (safely!). */ static int @@ -2173,6 +2234,15 @@ main(int argc, char **argv) io_nanny_start(ch_user, ctx.qc_tko * ctx.qc_interval); if (quorum_loop(&ctx, ni, MAX_NODES_DISK) == 0) { + /* + * if we are master and we are in master-win mode, + * request other qdiskd to elect a new one + */ + if ((ctx.qc_status == S_MASTER) && + ((ctx.qc_flags & RF_MASTER_WINS) || + (ctx.qc_flags & RF_AUTO_MASTER_WINS))) { + quorum_reelect_master(&ctx, ni, MAX_NODES_DISK); + } /* Only clean up if we're exiting w/o error) */ logt_print(LOG_NOTICE, "Unregistering quorum device.\n"); cman_unregister_quorum_device(ctx.qc_cman_admin);

1 0

cluster: RHEL59 - cman init: allow dlm hash table sizes to be tunable at startup
by Fabio M. Di Nitto 24 Jul '12

24 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=332db7754d621… Commit: 332db7754d6210991c34be056789aa1b33b26a0a Parent: 77bb92ad9dbad9c8755e45211214ab1e099cdb0e Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 08:55:27 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 08:55:27 2012 +0200 cman init: allow dlm hash table sizes to be tunable at startup Resolves: rhbz#836963 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/init.d/cman | 43 ++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 42 insertions(+), 1 deletions(-) diff --git a/cman/init.d/cman b/cman/init.d/cman index 4727d22..2101eb7 100755 --- a/cman/init.d/cman +++ b/cman/init.d/cman @@ -297,6 +297,37 @@ fence_join_enabled() fi } +tune_dlm_hash_sizes() +{ + dlmdir="/sys/kernel/config/dlm/cluster" + maxloop=21 + + if [ -z "$DLM_LKBTBL_SIZE" ] && \ + [ -z "$DLM_RSBTBL_SIZE" ] && \ + [ -z "$DLM_DIRTBL_SIZE" ]; then + return 0 + fi + for i in $(seq 1 $maxloop); do + if [ -d $dlmdir ]; then + break + fi + sleep 0.5 + done + if [ "$i" = "$maxloop" ]; then + return 1 + fi + if [ -n "$DLM_LKBTBL_SIZE" ]; then + echo $DLM_LKBTBL_SIZE > $dlmdir/lkbtbl_size + fi + if [ -n "$DLM_RSBTBL_SIZE" ]; then + echo $DLM_RSBTBL_SIZE > $dlmdir/rsbtbl_size + fi + if [ -n "$DLM_DIRTBL_SIZE" ]; then + echo $DLM_DIRTBL_SIZE > $dlmdir/dirtbl_size + fi + return 0 +} + start() { echo "Starting cluster: " @@ -391,7 +422,17 @@ start() return 1 fi fi - + + echo -n " Tuning DLM... " + tune_dlm_hash_sizes + if [ $? -eq 0 ] + then + echo "done" + else + echo "failed" + return 1 + fi + return 0 }

1 0

cluster: RHEL59 - fenced: fix double free when second method has no devices
by David Teigland 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=77bb92ad9dbad… Commit: 77bb92ad9dbad9c8755e45211214ab1e099cdb0e Parent: 6c9717cbbca38ee94911cbbc481c310f6375d560 Author: David Teigland <teigland(a)redhat.com> AuthorDate: Mon Jul 23 17:01:57 2012 -0500 Committer: David Teigland <teigland(a)redhat.com> CommitterDate: Mon Jul 23 17:01:57 2012 -0500 fenced: fix double free when second method has no devices When the second fence method is empty, the already freed device is freed again, causing a segfault. bz 809390 Signed-off-by: David Teigland <teigland(a)redhat.com> --- fence/fenced/agent.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/fence/fenced/agent.c b/fence/fenced/agent.c index 753b915..de4467c 100644 --- a/fence/fenced/agent.c +++ b/fence/fenced/agent.c @@ -361,8 +361,10 @@ int dispatch_fence_agent(char *victim, int force) device = NULL; } - if (device) + if (device) { free(device); + device = NULL; + } free(method); if (!error) {

1 0

cluster: RHEL59 - fenced: increase MAX_DEVICES to 8
by David Teigland 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=6c9717cbbca38… Commit: 6c9717cbbca38ee94911cbbc481c310f6375d560 Parent: b5253b0c74175ecda233f573fb0afa733b60eda2 Author: David Teigland <teigland(a)redhat.com> AuthorDate: Mon Jul 23 10:23:56 2012 -0500 Committer: David Teigland <teigland(a)redhat.com> CommitterDate: Mon Jul 23 16:36:44 2012 -0500 fenced: increase MAX_DEVICES to 8 Make this match rhel6. bz 821857 Signed-off-by: David Teigland <teigland(a)redhat.com> --- fence/fenced/agent.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/fence/fenced/agent.c b/fence/fenced/agent.c index e3aab9a..753b915 100644 --- a/fence/fenced/agent.c +++ b/fence/fenced/agent.c @@ -27,7 +27,7 @@ #include "ccs.h" #define MAX_METHODS 8 -#define MAX_DEVICES 4 +#define MAX_DEVICES 8 #define MAX_AGENT_ARGS_LEN 512 #define METHOD_NAME_PATH "/cluster/clusternodes/clusternode[@name=\"%s\"]/fence/method[%d]/@name"

1 0

cluster: RHEL59 - fencing: Automatic detection of EOL during login process
by Marek Grác 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=b5253b0c74175… Commit: b5253b0c74175ecda233f573fb0afa733b60eda2 Parent: 7d64a32254aac1c4bbeb7d7d684c606893d9fda1 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Mon Jul 23 17:23:39 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 17:23:39 2012 +0200 fencing: Automatic detection of EOL during login process Resolves: rhbz#810949 --- fence/agents/apc/fence_apc.py | 44 +++++++++++----------- fence/agents/bladecenter/fence_bladecenter.py | 18 +++++----- fence/agents/drac/fence_drac5.py | 12 +++--- fence/agents/ilo_mp/fence_ilo_mp.py | 12 +++--- fence/agents/lib/fencing.py.py | 48 +++++++++++++++--------- fence/agents/rsa/fence_rsa.py | 6 ++-- fence/agents/sanbox2/fence_sanbox2.py | 32 ++++++++-------- fence/agents/wti/fence_wti.py | 2 +- 8 files changed, 93 insertions(+), 81 deletions(-) diff --git a/fence/agents/apc/fence_apc.py b/fence/agents/apc/fence_apc.py index 0a3ce58..4125d39 100755 --- a/fence/agents/apc/fence_apc.py +++ b/fence/agents/apc/fence_apc.py @@ -29,7 +29,7 @@ def get_power_status(conn, options): exp_result = 0 outlets = {} try: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) version = 0 @@ -58,15 +58,15 @@ def get_power_status(conn, options): if switch == 0: if version == 2: if admin == 0: - conn.send("2\r\n") + conn.send_eol("2") else: - conn.send("3\r\n") + conn.send_eol("3") else: - conn.send("2\r\n") + conn.send_eol("2") conn.log_expect(options, options["-c"], int(options["-Y"])) - conn.send("1\r\n") + conn.send_eol("1") else: - conn.send(options["-s"]+"\r\n") + conn.send_eol(options["-s"]) while True: exp_result = conn.log_expect(options, [ options["-c"], "Press <ENTER>" ], int(options["-Y"])) @@ -76,7 +76,7 @@ def get_power_status(conn, options): res = show_re.search(x) if (res != None): outlets[res.group(2)] = (res.group(3), res.group(4)) - conn.send("\r\n") + conn.send_eol("") if exp_result == 0: break conn.send(chr(03)) @@ -103,7 +103,7 @@ def set_power_status(conn, options): }[options["-o"]] try: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) version = 0 @@ -138,41 +138,41 @@ def set_power_status(conn, options): if switch == 0: if version == 2: if admin2 == 0: - conn.send("2\r\n") + conn.send_eol("2") else: - conn.send("3\r\n") + conn.send_eol("3") else: - conn.send("2\r\n") + conn.send_eol("2") conn.log_expect(options, options["-c"], int(options["-Y"])) if (None == re.compile('.*2- Outlet Restriction.*', re.IGNORECASE | re.S).match(conn.before)): admin3 = 0 else: admin3 = 1 - conn.send("1\r\n") + conn.send_eol("1") else: - conn.send(options["-s"] + "\r\n") + conn.send_eol(options["-s"]) while 1 == conn.log_expect(options, [ options["-c"], "Press <ENTER>" ], int(options["-Y"])): - conn.send("\r\n") - conn.send(options["-n"]+"\r\n") + conn.send_eol("") + conn.send_eol(options["-n"]+"") conn.log_expect(options, options["-c"], int(options["-Y"])) if switch == 0: if admin2 == 1: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) if admin3 == 1: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) else: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) - conn.send(action+"\r\n") + conn.send_eol(action) conn.log_expect(options, "Enter 'YES' to continue or <ENTER> to cancel :", int(options["-Y"])) - conn.send("YES\r\n") + conn.send_eol("YES") conn.log_expect(options, "Press <ENTER> to continue...", int(options["-Y"])) - conn.send("\r\n") + conn.send_eol("") conn.log_expect(options, options["-c"], int(options["-Y"])) conn.send(chr(03)) conn.log_expect(options, "- Logout", int(options["-Y"])) @@ -231,7 +231,7 @@ will block any necessary fencing actions." ## a problem because everything is checked before. ###### try: - conn.sendline("4") + conn.send_eol("4") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/bladecenter/fence_bladecenter.py b/fence/agents/bladecenter/fence_bladecenter.py index ac65f65..d97b9e4 100755 --- a/fence/agents/bladecenter/fence_bladecenter.py +++ b/fence/agents/bladecenter/fence_bladecenter.py @@ -27,7 +27,7 @@ def get_power_status(conn, options): try: node_cmd = "system:blade\[" + options["-n"] + "\]>" - conn.send("env -T system:blade[" + options["-n"] + "]\r\n") + conn.send_eol("env -T system:blade[" + options["-n"] + "]") i = conn.log_expect(options, [ node_cmd, "system>" ] , int(options["-Y"])) if i == 1: ## Given blade number does not exist @@ -35,10 +35,10 @@ def get_power_status(conn, options): return "off" else: fail(EC_STATUS) - conn.send("power -state\r\n") + conn.send_eol("power -state") conn.log_expect(options, node_cmd, int(options["-Y"])) status = conn.before.splitlines()[-1] - conn.send("env -T system\r\n") + conn.send_eol("env -T system") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -56,7 +56,7 @@ def set_power_status(conn, options): try: node_cmd = "system:blade\[" + options["-n"] + "\]>" - conn.send("env -T system:blade[" + options["-n"] + "]\r\n") + conn.send_eol("env -T system:blade[" + options["-n"] + "]") i = conn.log_expect(options, [ node_cmd, "system>" ] , int(options["-Y"])) if i == 1: ## Given blade number does not exist @@ -65,9 +65,9 @@ def set_power_status(conn, options): else: fail(EC_GENERIC_ERROR) - conn.send("power -"+options["-o"]+"\r\n") + conn.send_eol("power -"+options["-o"]) conn.log_expect(options, node_cmd, int(options["-Y"])) - conn.send("env -T system\r\n") + conn.send_eol("env -T system") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -79,9 +79,9 @@ def get_blades_list(conn, options): try: node_cmd = "system>" - conn.send("env -T system\r\n") + conn.send_eol("env -T system") conn.log_expect(options, node_cmd, int(options["-Y"])) - conn.send("list -l 2\r\n") + conn.send_eol("list -l 2") conn.log_expect(options, node_cmd, int(options["-Y"])) lines = conn.before.split("\r\n") @@ -131,7 +131,7 @@ and uses the command line interface to power on and off blades." ## Logout from system ###### try: - conn.send("exit\r\n") + conn.send_eol("exit") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/drac/fence_drac5.py b/fence/agents/drac/fence_drac5.py index 23cd7ca..298339f 100755 --- a/fence/agents/drac/fence_drac5.py +++ b/fence/agents/drac/fence_drac5.py @@ -25,9 +25,9 @@ BUILD_DATE="" def get_power_status(conn, options): try: if options["model"] == "DRAC CMC": - conn.sendline("racadm serveraction powerstatus -m " + options["-m"]) + conn.send_eol("racadm serveraction powerstatus -m " + options["-m"]) elif options["model"] == "DRAC 5": - conn.sendline("racadm serveraction powerstatus") + conn.send_eol("racadm serveraction powerstatus") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: @@ -49,9 +49,9 @@ def set_power_status(conn, options): try: if options["model"] == "DRAC CMC": - conn.sendline("racadm serveraction " + action + " -m " + options["-m"]) + conn.send_eol("racadm serveraction " + action + " -m " + options["-m"]) elif options["model"] == "DRAC 5": - conn.sendline("racadm serveraction " + action) + conn.send_eol("racadm serveraction " + action) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -63,7 +63,7 @@ def get_list_devices(conn, options): try: if options["model"] == "DRAC CMC": - conn.sendline("getmodinfo") + conn.send_eol("getmodinfo") list_re = re.compile("^([^\s]*?)\s+Present\s*(ON|OFF)\s*.*$") conn.log_expect(options, options["-c"], int(options["-g"])) @@ -133,7 +133,7 @@ By default, the telnet interface is not enabled." ## Logout from system ###### try: - conn.sendline("exit") + conn.send_eol("exit") time.sleep(1) conn.close() except exceptions.OSError: diff --git a/fence/agents/ilo_mp/fence_ilo_mp.py b/fence/agents/ilo_mp/fence_ilo_mp.py index 06e1358..40ac597 100644 --- a/fence/agents/ilo_mp/fence_ilo_mp.py +++ b/fence/agents/ilo_mp/fence_ilo_mp.py @@ -12,7 +12,7 @@ BUILD_DATE="" def get_power_status(conn, options): try: - conn.send("show /system1\r\n") + conn.send_eol("show /system1") re_state = re.compile('EnabledState=(.*)', re.IGNORECASE) conn.log_expect(options, re_state, int(options["-Y"])) @@ -31,9 +31,9 @@ def get_power_status(conn, options): def set_power_status(conn, options): try: if options["-o"] == "on": - conn.send("start /system1\r\n") + conn.send_eol("start /system1") else: - conn.send("stop -f /system1\r\n") + conn.send_eol("stop -f /system1") conn.log_expect(options, options["-c"], int(options["-g"])) @@ -46,7 +46,7 @@ def set_power_status(conn, options): def main(): device_opt = [ "help", "version", "agent", "quiet", "verbose", "debug", "action", "ipaddr", "login", "passwd", "passwd_script", - "secure", "identity_file", "cmd_prompt", "ipport", "login_eol_lf", + "secure", "identity_file", "cmd_prompt", "ipport", "separator", "inet4_only", "inet6_only", "power_timeout", "shell_timeout", "login_timeout", "power_wait" ] @@ -64,7 +64,7 @@ def main(): show_docs(options, docs) conn = fence_login(options) - conn.send("SMCLP\r\n") + conn.send_eol("SMCLP") ## ## Fence operations @@ -72,7 +72,7 @@ def main(): result = fence_action(conn, options, set_power_status, get_power_status) try: - conn.send("exit\r\n") + conn.send_eol("exit") except exceptions.OSError: pass except pexpect.ExceptionPexpect: diff --git a/fence/agents/lib/fencing.py.py b/fence/agents/lib/fencing.py.py index c4fe218..25a0f7e 100644 --- a/fence/agents/lib/fencing.py.py +++ b/fence/agents/lib/fencing.py.py @@ -178,11 +178,6 @@ all_opt = { "required" : "0", "shortdesc" : "Force ribcl version to use", "order" : 1 }, - "login_eol_lf" : { - "getopt" : "", - "help" : "", - "order" : 1 - }, "cmd_prompt" : { "getopt" : "c:", "longopt" : "command-prompt", @@ -403,12 +398,20 @@ all_opt = { common_opt = [ "retry_on", "delay" ] class fspawn(pexpect.spawn): + def __init__(self, options, command): + pexpect.spawn.__init__(self, command) + self.opt = options + def log_expect(self, options, pattern, timeout): result = self.expect(pattern, timeout) if options["log"] >= LOG_MODE_VERBOSE: options["debug_fh"].write(self.before + self.after) return result + # send EOL according to what was detected in login process (telnet) + def send_eol(self, message): + self.send(message + self.opt["eol"]) + def atexit_handler(): try: sys.stdout.close() @@ -859,10 +862,7 @@ def fence_login(options): if (options.has_key("-4")): force_ipvx="-4 " - if (options["device_opt"].count("login_eol_lf")): - login_eol = "\n" - else: - login_eol = "\r\n" + options["eol"] = "\r\n" ## Do the delay of the fence device before logging in ## Delay is important for two-node clusters fencing but we do not need to delay 'status' operations @@ -876,7 +876,7 @@ def fence_login(options): if options.has_key("-z"): command = '%s %s %s %s' % (SSL_PATH, force_ipvx, options["-a"], options["-u"]) try: - conn = fspawn(command) + conn = fspawn(options, command) except pexpect.ExceptionPexpect, ex: ## SSL telnet is part of the fencing package sys.stderr.write(str(ex) + "\n") @@ -886,7 +886,7 @@ def fence_login(options): if options.has_key("ssh_options"): command += ' ' + options["ssh_options"] try: - conn = fspawn(command) + conn = fspawn(options, command) except pexpect.ExceptionPexpect, ex: sys.stderr.write(str(ex) + "\n") sys.stderr.write("Due to limitations, binary dependencies on fence agents " @@ -915,7 +915,7 @@ def fence_login(options): if options.has_key("ssh_options"): command += ' ' + options["ssh_options"] try: - conn = fspawn(command) + conn = fspawn(options, command) except pexpect.ExceptionPexpect, ex: sys.stderr.write(str(ex) + "\n") sys.stderr.write("Due to limitations, binary dependencies on fence agents " @@ -934,7 +934,7 @@ def fence_login(options): fail_usage("Failed: You have to enter passphrase (-p) for identity file") else: try: - conn = fspawn(TELNET_PATH) + conn = fspawn(options, TELNET_PATH) conn.send("set binary\n") conn.send("open %s -%s\n"%(options["-a"], options["-u"])) except pexpect.ExceptionPexpect, ex: @@ -943,11 +943,23 @@ def fence_login(options): "are not in the spec file and must be installed separately." + "\n") sys.exit(EC_GENERIC_ERROR) - conn.log_expect(options, re_login, int(options["-y"])) - conn.send(options["-l"] + login_eol) - conn.log_expect(options, re_pass, int(options["-Y"])) - conn.send(options["-p"] + login_eol) - conn.log_expect(options, options["-c"], int(options["-Y"])) + result = conn.log_expect(options, re_login, int(options["-y"])) + conn.send_eol(options["-l"]) + + ## automatically change end of line separator + screen = conn.read_nonblocking(size=100, timeout=int(options["-Y"])) + if (re_login.search(screen) != None): + options["eol"] = "\n" + conn.send_eol(options["-l"]) + result = conn.log_expect(options, re_pass, int(options["-y"])) + elif (re_pass.search(screen) == None): + conn.log_expect(options, re_pass, int(options["-Y"])) + + try: + conn.send_eol(options["-p"]) + conn.log_expect(options, options["-c"], int(options["-Y"])) + except KeyError: + fail(EC_PASSWORD_MISSING) except pexpect.EOF: fail(EC_LOGIN_DENIED) except pexpect.TIMEOUT: diff --git a/fence/agents/rsa/fence_rsa.py b/fence/agents/rsa/fence_rsa.py index 33c7fe9..d82c448 100755 --- a/fence/agents/rsa/fence_rsa.py +++ b/fence/agents/rsa/fence_rsa.py @@ -19,7 +19,7 @@ BUILD_DATE="March, 2009" def get_power_status(conn, options): try: - conn.send("power state\r\n") + conn.send_eol("power state") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -36,7 +36,7 @@ def get_power_status(conn, options): def set_power_status(conn, options): try: - conn.send("power " + options["-o"] + "\r\n") + conn.send_eol("power " + options["-o"]) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -84,7 +84,7 @@ will block any necessary fencing actions." ## Logout from system ###### try: - conn.sendline("exit") + conn.send_eol("exit") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/sanbox2/fence_sanbox2.py b/fence/agents/sanbox2/fence_sanbox2.py index c95e602..d20f7e2 100644 --- a/fence/agents/sanbox2/fence_sanbox2.py +++ b/fence/agents/sanbox2/fence_sanbox2.py @@ -24,14 +24,14 @@ def get_power_status(conn, options): 'offline' : "off" } try: - conn.send("show port " + options["-n"] + "\n") + conn.send_eol("show port " + options["-n"]) conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass @@ -51,28 +51,28 @@ def set_power_status(conn, options): }[options["-o"]] try: - conn.send("set port " + options["-n"] + " state " + action + "\n") + conn.send_eol("set port " + options["-n"] + " state " + action) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass fail(EC_TIMED_OUT) try: - conn.send("set port " + options["-n"] + " state " + action + "\n") + conn.send_eol("set port " + options["-n"] + " state " + action) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass @@ -82,7 +82,7 @@ def get_list_devices(conn, options): outlets = { } try: - conn.send("show port" + "\n") + conn.send_eol("show port") conn.log_expect(options, options["-c"], int(options["-Y"])) list_re = re.compile("^\s+(\d+?)\s+(Online|Offline)\s+", re.IGNORECASE) @@ -98,8 +98,8 @@ def get_list_devices(conn, options): fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass @@ -110,7 +110,7 @@ def get_list_devices(conn, options): def main(): device_opt = [ "help", "version", "agent", "quiet", "verbose", "debug", "io_fencing", "ipaddr", "login", "passwd", "passwd_script", - "cmd_prompt", "port", "ipport", "login_eol_lf", "separator", + "cmd_prompt", "port", "ipport", "separator", "power_timeout", "shell_timeout", "login_timeout", "power_wait" ] atexit.register(atexit_handler) @@ -138,7 +138,7 @@ because the connection will block any necessary fencing actions." ## conn = fence_login(options) - conn.send("admin start\n") + conn.send_eol("admin start") conn.log_expect(options, options["-c"], int(options["-Y"])) if (re.search("$admin$", conn.before, re.MULTILINE) == None): @@ -153,8 +153,8 @@ because the connection will block any necessary fencing actions." ## Logout from system ###### try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit\n") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/wti/fence_wti.py b/fence/agents/wti/fence_wti.py index c742e71..40b97e3 100755 --- a/fence/agents/wti/fence_wti.py +++ b/fence/agents/wti/fence_wti.py @@ -125,7 +125,7 @@ is running because the connection will block any necessary fencing actions." if 0 == options.has_key("-x"): try: try: - conn = fspawn(TELNET_PATH) + conn = fspawn(options, TELNET_PATH) conn.send("set binary\n") conn.send("open %s -%s\n"%(options["-a"], options["-u"])) except pexpect.ExceptionPexpect, ex:

1 0

cluster: RHEL59 - fence_apc: fence_apc fails for some port numbers
by Marek Grác 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=7d64a32254aac… Commit: 7d64a32254aac1c4bbeb7d7d684c606893d9fda1 Parent: 1ed8b72936fa7d871dbdc9b787e7e775d1ea3699 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Wed Jun 23 18:29:02 2010 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 17:20:28 2012 +0200 fence_apc: fence_apc fails for some port numbers Problem is that string we parse looks like: 'Press Enter to continue ...\r 23-bar1 ON' and we did not expect text message before our line. Resolves: #606315 --- fence/agents/apc/fence_apc.py | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/fence/agents/apc/fence_apc.py b/fence/agents/apc/fence_apc.py index 1aa7bed..0a3ce58 100755 --- a/fence/agents/apc/fence_apc.py +++ b/fence/agents/apc/fence_apc.py @@ -71,11 +71,11 @@ def get_power_status(conn, options): while True: exp_result = conn.log_expect(options, [ options["-c"], "Press <ENTER>" ], int(options["-Y"])) lines = conn.before.split("\n"); - show_re = re.compile('^\s*(\d+)- (.*?)\s+(ON|OFF)\s*') + show_re = re.compile('(^|\x0D)\s*(\d+)- (.*?)\s+(ON|OFF)\s*') for x in lines: res = show_re.search(x) if (res != None): - outlets[res.group(1)] = (res.group(2), res.group(3)) + outlets[res.group(2)] = (res.group(3), res.group(4)) conn.send("\r\n") if exp_result == 0: break

1 0

cluster: RHEL59 - fencing: Add iPDU fencing agent for model 46M4002
by Marek Grác 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=1ed8b72936fa7… Commit: 1ed8b72936fa7d871dbdc9b787e7e775d1ea3699 Parent: 76741bb2a94ae94e493c609d50f570d02e2f3029 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Mon Jul 23 17:06:32 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 17:06:32 2012 +0200 fencing: Add iPDU fencing agent for model 46M4002 Resolves: rhbz#741985 --- fence/agents/Makefile | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/fence/agents/Makefile b/fence/agents/Makefile index 716530c..669e291 100644 --- a/fence/agents/Makefile +++ b/fence/agents/Makefile @@ -30,6 +30,7 @@ all: ${MAKE} -C ifmib all ${MAKE} -C ilo all ${MAKE} -C ilo_mp all + ${MAKE} -C ipdu all ${MAKE} -C ipmilan all ${MAKE} -C lpar all ${MAKE} -C manual all @@ -65,6 +66,7 @@ install: all ${MAKE} -C ifmib install ${MAKE} -C ilo install ${MAKE} -C ilo_mp install + ${MAKE} -C ipdu install ${MAKE} -C ipmilan install ${MAKE} -C lpar install ${MAKE} -C manual install @@ -100,6 +102,7 @@ clean: ${MAKE} -C ifmib clean ${MAKE} -C ilo clean ${MAKE} -C ilo_mp clean + ${MAKE} -C ipdu clean ${MAKE} -C ipmilan clean ${MAKE} -C lpar clean ${MAKE} -C manual clean

1 0

cluster: RHEL59 - qdiskd: Make multipath issues go away
by Fabio M. Di Nitto 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=76741bb2a94ae… Commit: 76741bb2a94ae94e493c609d50f570d02e2f3029 Parent: 3290dff9b88992e913b28dd57c6b64a3c2b75c8a Author: Lon Hohberger <lhh(a)redhat.com> AuthorDate: Fri May 6 10:14:04 2011 -0400 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Mon Jul 23 14:54:07 2012 +0200 qdiskd: Make multipath issues go away Qdiskd hsitorically has required significant tuning to work around delays which occur during multipath failover, overloaded I/O, and LUN trespasses in both device-mapper-multipath and EMC PowerPath environments. This patch goes a very long way towards eliminating false evictions when these conditions occur by making qdiskd whine to the other cluster members when it detects hung system calls. When a cluster member whines, it indicates the source of the problem (which system call is hung), and the act of receiving a whine from a host indicates that qdiskd is operational, but that I/O is hung. Hung I/O is different from losing storage entirely (where you get I/O errors). Possible problems: - Receive queue getting very full, causing messages to become blocked on a node where I/O is hung. 1) that would take a very long time, and 2) node should get evicted at that point anyway. Resolves: rhbz#782900 this version of the patch is a backport of: e2937eb33f224f86904fead08499a6178868ca6a 34d2872fb7e60be1594158acaaeb8acd74f78d22 There is a minor change vs original patch based on how qdiskd in RHEL5 handles cman connection. We add an extra call to cman_alive in main qdisk_loop to make sure data are not stalled on the cman port, and data_callback to qdiskd_whine executed. Signed-off-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/daemon/cnxman-socket.h | 1 + cman/qdisk/Makefile | 2 +- cman/qdisk/disk.h | 6 ++++ cman/qdisk/iostate.c | 17 +++++++++++-- cman/qdisk/iostate.h | 4 ++- cman/qdisk/main.c | 54 +++++++++++++++++++++++++++++++++++++++---- 6 files changed, 74 insertions(+), 10 deletions(-) diff --git a/cman/daemon/cnxman-socket.h b/cman/daemon/cnxman-socket.h index 351c97c..1d01b44 100644 --- a/cman/daemon/cnxman-socket.h +++ b/cman/daemon/cnxman-socket.h @@ -79,6 +79,7 @@ #define CLUSTER_PORT_SERVICES 2 #define CLUSTER_PORT_SYSMAN 10 /* Remote execution daemon */ #define CLUSTER_PORT_CLVMD 11 /* Cluster LVM daemon */ +#define CLUSTER_PORT_QDISKD 178 /* Quorum disk daemon */ /* Port numbers above this will be blocked when the cluster is inquorate or in * transition */ diff --git a/cman/qdisk/Makefile b/cman/qdisk/Makefile index f58806b..9bfc486 100644 --- a/cman/qdisk/Makefile +++ b/cman/qdisk/Makefile @@ -32,7 +32,7 @@ qdiskd: disk.o crc32.o disk_util.o main.o score.o bitmap.o clulog.o \ gcc -o $@ $^ -lpthread -L../lib -L${ccslibdir} -lccs -lrt mkqdisk: disk.o crc32.o disk_util.o iostate.o \ - proc.o mkqdisk.o scandisk.o clulog.o gettid.o + proc.o mkqdisk.o scandisk.o clulog.o gettid.o ../lib/libcman.a gcc -o $@ $^ -lrt %.o: %.c diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index b784220..d491de1 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -290,6 +290,12 @@ typedef struct { status_block_t ni_status; } node_info_t; +typedef struct { + qd_ctx *ctx; + node_info_t *ni; + size_t ni_len; +} qd_priv_t; + int qd_write_status(qd_ctx *ctx, int nid, disk_node_state_t state, disk_msg_t *msg, memb_mask_t mask, memb_mask_t master); int qd_read_print_status(target_info_t *disk, int nid); diff --git a/cman/qdisk/iostate.c b/cman/qdisk/iostate.c index 65b4d50..eb74ad2 100644 --- a/cman/qdisk/iostate.c +++ b/cman/qdisk/iostate.c @@ -1,10 +1,14 @@ #include <pthread.h> +#include <libcman.h> #include <iostate.h> #include <unistd.h> #include <time.h> #include <sys/time.h> #include <clulog.h> +#include <stdint.h> +#include "platform.h" #include "iostate.h" +#include "../daemon/cnxman-socket.h" static iostate_t main_state = 0; static int main_incarnation = 0; @@ -26,7 +30,7 @@ static struct state_table io_state_table[] = { { STATE_LSEEK, "seek" }, { -1, NULL } }; -static const char * +const char * state_to_string(iostate_t state) { static const char *ret = "unknown"; @@ -65,6 +69,8 @@ io_nanny_thread(void *arg) iostate_t last_main_state = 0, current_main_state = 0; int last_main_incarnation = 0, current_main_incarnation = 0; int logged_incarnation = 0; + cman_handle_t ch = (cman_handle_t)arg; + int32_t whine_state; /* Start with wherever we're at now */ pthread_mutex_lock(&state_mutex); @@ -96,6 +102,11 @@ io_nanny_thread(void *arg) continue; } + /* Whine on CMAN api */ + whine_state = (int32_t)current_main_state; + swab32(whine_state); + cman_send_data(ch, &whine_state, sizeof(int32_t), 0, CLUSTER_PORT_QDISKD, 0); + /* Don't log things twice */ if (logged_incarnation == current_main_incarnation) continue; @@ -114,7 +125,7 @@ io_nanny_thread(void *arg) int -io_nanny_start(int timeout) +io_nanny_start(cman_handle_t ch, int timeout) { int ret; @@ -124,7 +135,7 @@ io_nanny_start(int timeout) qdisk_timeout = timeout; thread_active = 1; - ret = pthread_create(&io_nanny_tid, NULL, io_nanny_thread, NULL); + ret = pthread_create(&io_nanny_tid, NULL, io_nanny_thread, ch); pthread_mutex_unlock(&state_mutex); return ret; diff --git a/cman/qdisk/iostate.h b/cman/qdisk/iostate.h index 7dd7bf6..a65b1d4 100644 --- a/cman/qdisk/iostate.h +++ b/cman/qdisk/iostate.h @@ -11,7 +11,9 @@ typedef enum { void io_state(iostate_t state); -int io_nanny_start(int timeout); +int io_nanny_start(cman_handle_t ch, int timeout); int io_nanny_stop(void); +const char * state_to_string(iostate_t state); + #endif diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index 0d7bb3d..90d00ab 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -48,6 +48,7 @@ (defined(LIBCMAN_VERSION) && LIBCMAN_VERSION < 2)) #include <cluster/cnxman-socket.h> #endif +#include "../daemon/cnxman-socket.h" int daemon_init(char *); int check_process_running(char *, pid_t *); @@ -892,6 +893,11 @@ quorum_loop(qd_ctx *ctx, node_info_t *ni, int max) _running = 1; while (_running) { + /* perform a forceful cman dispatch */ + if (cman_alive(ctx->qc_ch) < 0) { + clulog(LOG_ERR, "cman: %s\n", strerror(errno)); + } + /* XXX this was getuptime() in clumanager */ get_time(&oldtime, (ctx->qc_flags&RF_UPTIME)); @@ -1514,6 +1520,31 @@ check_stop_cman(qd_ctx *ctx) } } +static void +qdisk_whine(cman_handle_t h, void *privdata, char *buf, int len, + uint8_t port, int nodeid) +{ + int32_t dstate; + qd_priv_t *qp = (qd_priv_t *)privdata; + node_info_t *ni = qp->ni; + + if (len != sizeof(dstate)) { + return; + } + + dstate = *((int32_t*)buf); + + if (nodeid == (qp->ctx->qc_my_id)) + return; + + swab32(dstate); + + if (dstate) { + clulog(LOG_CRIT, "qdiskd on node %d reports hung %s()\n", + state_to_string(dstate)); + ni[nodeid-1].ni_misses = 0; + } +} int main(int argc, char **argv) @@ -1528,6 +1559,7 @@ main(int argc, char **argv) char device[128]; pid_t pid; quorum_header_t qh; + qd_priv_t qp; if (check_process_running(argv[0], &pid) && pid !=getpid()) { printf("QDisk services already running\n"); @@ -1559,10 +1591,16 @@ main(int argc, char **argv) } } + /* For cman notifications we need two sockets - one for events, + one for config change callbacks */ + qp.ctx = &ctx; + qp.ni = &ni[0]; + qp.ni_len = MAX_NODES_DISK; + #if (defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2) - ch = cman_admin_init(NULL); + ch = cman_admin_init(&qp); #else - ch = cman_init(NULL); + ch = cman_init(&qp); #endif if (!ch) { if (!foreground && !forked) { @@ -1577,13 +1615,19 @@ main(int argc, char **argv) do { sleep(5); #if (defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2) - ch = cman_admin_init(NULL); + ch = cman_admin_init(&qp); #else - ch = cman_init(NULL); + ch = cman_init(&qp); #endif } while (!ch); } + if (cman_start_recv_data(ch, qdisk_whine, CLUSTER_PORT_QDISKD) != 0) { + clulog_and_print(LOG_CRIT, "Could not register with CMAN: %s\n", + strerror(errno)); + goto out; + } + memset(&me, 0, sizeof(me)); while (cman_get_node(ch, CMAN_NODEID_US, &me) < 0) { if (!foreground && !forked) { @@ -1696,7 +1740,7 @@ main(int argc, char **argv) } } - io_nanny_start(ctx.qc_tko * ctx.qc_interval); + io_nanny_start(ch, ctx.qc_tko * ctx.qc_interval); if (quorum_loop(&ctx, ni, MAX_NODES_DISK) == 0) cman_unregister_quorum_device(ctx.qc_ch);

1 0

fence-agents: master - fence_ipdu: Minor fixes to fence agent
by Marek Grác 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=dff6b12a… Commit: dff6b12ab73c8c38fc579ec312f398d6711cfeb8 Parent: 74d1242d8b968d5886be4394c8792d8c8d93f321 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Mon Jul 23 11:24:43 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 11:24:43 2012 +0200 fence_ipdu: Minor fixes to fence agent * replace code to use 'default' value * remove code for backward compatibility with old options (-n switch:port) Resolves: rhbz#741985 --- fence/agents/ipdu/fence_ipdu.py | 13 ++----------- 1 files changed, 2 insertions(+), 11 deletions(-) diff --git a/fence/agents/ipdu/fence_ipdu.py b/fence/agents/ipdu/fence_ipdu.py index 9c135e6..aac0b7d 100644 --- a/fence/agents/ipdu/fence_ipdu.py +++ b/fence/agents/ipdu/fence_ipdu.py @@ -49,7 +49,7 @@ def ipdu_set_device(conn,options): agents_dir={'.1.3.6.1.4.1.2.6.223':IBMiPDU, None:IBMiPDU} - # First resolve type of APC + # First resolve type of PDU device pdu_type=conn.walk(OID_SYS_OBJECT_ID) if (not ((len(pdu_type)==1) and (agents_dir.has_key(pdu_type[0][1])))): @@ -133,6 +133,7 @@ def get_outlets_status(conn, options): def ipdu_snmp_define_defaults(): all_opt["snmp_version"]["default"]="3" all_opt["community"]["default"]="private" + all_opt["switch"]["default"]="1" device=IBMiPDU # Main agent method @@ -152,16 +153,6 @@ def main(): options=check_input(device_opt,process_input(device_opt)) - ## Support for -n [switch]:[plug] notation that was used before - if ((options.has_key("-n")) and (-1 != options["-n"].find(":"))): - (switch, plug) = options["-n"].split(":", 1) - if ((switch.isdigit()) and (plug.isdigit())): - options["-s"] = switch - options["-n"] = plug - - if (not (options.has_key("-s"))): - options["-s"]="1" - docs = { } docs["shortdesc"] = "Fence agent for iPDU over SNMP" docs["longdesc"] = "fence_ipdu is an I/O Fencing agent \

1 0

fence-agents: master - Add iPDU fencing agent for model 46M4002
by Marek Grác 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=74d1242d… Commit: 74d1242d8b968d5886be4394c8792d8c8d93f321 Parent: efcf34f5156385c397509f9fe40ae08ca74a96a0 Author: Lon Hohberger <lhh(a)redhat.com> AuthorDate: Fri Jul 20 15:43:14 2012 -0400 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 11:10:19 2012 +0200 Add iPDU fencing agent for model 46M4002 Resolves: rhbz#741985 Signed-off-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Marek 'marx' Grac <mgrac(a)redhat.com> --- .gitignore | 1 + configure.ac | 1 + fence/agents/ipdu/Makefile.am | 17 ++++ fence/agents/ipdu/fence_ipdu.py | 179 +++++++++++++++++++++++++++++++++++++++ 4 files changed, 198 insertions(+), 0 deletions(-) diff --git a/.gitignore b/.gitignore index 3a182d8..c1b028c 100644 --- a/.gitignore +++ b/.gitignore @@ -49,6 +49,7 @@ fence/agents/ifmib/fence_ifmib fence/agents/ilo/fence_ilo fence/agents/ilo_mp/fence_ilo_mp fence/agents/intelmodular/fence_intelmodular +fence/agents/ipdu/fence_ipdu fence/agents/ipmilan/fence_ipmilan fence/agents/ldom/fence_ldom fence/agents/lib/fencing.py diff --git a/configure.ac b/configure.ac index 4797740..1b0ee22 100644 --- a/configure.ac +++ b/configure.ac @@ -263,6 +263,7 @@ AC_CONFIG_FILES([Makefile fence/agents/eps/Makefile fence/agents/hpblade/Makefile fence/agents/ibmblade/Makefile + fence/agents/ipdu/Makefile fence/agents/ifmib/Makefile fence/agents/ilo/Makefile fence/agents/ilo_mp/Makefile diff --git a/fence/agents/ipdu/Makefile.am b/fence/agents/ipdu/Makefile.am new file mode 100644 index 0000000..fb02cc8 --- /dev/null +++ b/fence/agents/ipdu/Makefile.am @@ -0,0 +1,17 @@ +MAINTAINERCLEANFILES = Makefile.in + +TARGET = fence_ipdu + +SRC = $(TARGET).py + +EXTRA_DIST = $(SRC) + +sbin_SCRIPTS = $(TARGET) + +man_MANS = $(TARGET).8 + +include $(top_srcdir)/make/fencebuild.mk +include $(top_srcdir)/make/fenceman.mk + +clean-local: clean-man + rm -f $(TARGET) diff --git a/fence/agents/ipdu/fence_ipdu.py b/fence/agents/ipdu/fence_ipdu.py new file mode 100644 index 0000000..9c135e6 --- /dev/null +++ b/fence/agents/ipdu/fence_ipdu.py @@ -0,0 +1,179 @@ +#!/usr/bin/python + +# The Following agent has been tested on: +# IBM iPDU model 46M4002 +# Firmware release OPDP_sIBM_v01.2_1 +# + +import sys, re, pexpect +sys.path.append("/usr/share/fence") +from fencing import * +from fencing_snmp import * + +#BEGIN_VERSION_GENERATION +RELEASE_VERSION="IBM iPDU SNMP fence agent" +REDHAT_COPYRIGHT="" +BUILD_DATE="" +#END_VERSION_GENERATION + +### CONSTANTS ### +# oid defining fence device +OID_SYS_OBJECT_ID='.1.3.6.1.2.1.1.2.0' + +### GLOBAL VARIABLES ### +# Device - see IBM iPDU +device=None + +# Port ID +port_id=None +# Switch ID +switch_id=None + +# Classes describing Device params +class IBMiPDU: + # iPDU + status_oid= '.1.3.6.1.4.1.2.6.223.8.2.2.1.11.%d' + control_oid= '.1.3.6.1.4.1.2.6.223.8.2.2.1.11.%d' + outlet_table_oid='.1.3.6.1.4.1.2.6.223.8.2.2.1.2' + ident_str="IBM iPDU" + state_on=1 + state_off=0 + turn_on=1 + turn_off=0 + has_switches=False + +### FUNCTIONS ### +def ipdu_set_device(conn,options): + global device + + agents_dir={'.1.3.6.1.4.1.2.6.223':IBMiPDU, + None:IBMiPDU} + + # First resolve type of APC + pdu_type=conn.walk(OID_SYS_OBJECT_ID) + + if (not ((len(pdu_type)==1) and (agents_dir.has_key(pdu_type[0][1])))): + pdu_type=[[None,None]] + + device=agents_dir[pdu_type[0][1]] + + conn.log_command("Trying %s"%(device.ident_str)) + +def ipdu_resolv_port_id(conn,options): + global port_id,switch_id,device + + if (device==None): + ipdu_set_device(conn,options) + + # Now we resolv port_id/switch_id + if ((options["-n"].isdigit()) and ((not device.has_switches) or (options["-s"].isdigit()))): + port_id=int(options["-n"]) + + if (device.has_switches): + switch_id=int(options["-s"]) + else: + table=conn.walk(device.outlet_table_oid,30) + + for x in table: + if (x[1].strip('"')==options["-n"]): + t=x[0].split('.') + if (device.has_switches): + port_id=int(t[len(t)-1]) + switch_id=int(t[len(t)-3]) + else: + port_id=int(t[len(t)-1]) + + if (port_id==None): + fail_usage("Can't find port with name %s!"%(options["-n"])) + +def get_power_status(conn,options): + global port_id,switch_id,device + + if (port_id==None): + ipdu_resolv_port_id(conn,options) + + oid=((device.has_switches) and device.status_oid%(switch_id,port_id) or device.status_oid%(port_id)) + + (oid,status)=conn.get(oid) + return (status==str(device.state_on) and "on" or "off") + +def set_power_status(conn, options): + global port_id,switch_id,device + + if (port_id==None): + ipdu_resolv_port_id(conn,options) + + oid=((device.has_switches) and device.control_oid%(switch_id,port_id) or device.control_oid%(port_id)) + + conn.set(oid,(options["-o"]=="on" and device.turn_on or device.turn_off)) + + +def get_outlets_status(conn, options): + global device + + result={} + + if (device==None): + ipdu_set_device(conn,options) + + res_ports=conn.walk(device.outlet_table_oid,30) + + for x in res_ports: + t=x[0].split('.') + + port_num=((device.has_switches) and "%s:%s"%(t[len(t)-3],t[len(t)-1]) or "%s"%(t[len(t)-1])) + + port_name=x[1].strip('"') + port_status="" + result[port_num]=(port_name,port_status) + + return result + +# Define new options +def ipdu_snmp_define_defaults(): + all_opt["snmp_version"]["default"]="3" + all_opt["community"]["default"]="private" + device=IBMiPDU + +# Main agent method +def main(): + device_opt = [ "help", "version", "agent", "quiet", "verbose", "debug", + "action", "ipaddr", "login", "passwd", "passwd_script", + "test", "port", "separator", "no_login", "no_password", + "snmp_version", "community", "snmp_auth_prot", "snmp_sec_level", + "snmp_priv_prot", "snmp_priv_passwd", "snmp_priv_passwd_script", + "udpport","inet4_only","inet6_only", + "power_timeout", "shell_timeout", "login_timeout", "power_wait" ] + + atexit.register(atexit_handler) + + snmp_define_defaults () + ipdu_snmp_define_defaults() + + options=check_input(device_opt,process_input(device_opt)) + + ## Support for -n [switch]:[plug] notation that was used before + if ((options.has_key("-n")) and (-1 != options["-n"].find(":"))): + (switch, plug) = options["-n"].split(":", 1) + if ((switch.isdigit()) and (plug.isdigit())): + options["-s"] = switch + options["-n"] = plug + + if (not (options.has_key("-s"))): + options["-s"]="1" + + docs = { } + docs["shortdesc"] = "Fence agent for iPDU over SNMP" + docs["longdesc"] = "fence_ipdu is an I/O Fencing agent \ +which can be used with the IBM iPDU network power switch. It logs \ +into a device via SNMP and reboots a specified outlet. It supports \ +SNMP v3 with all combinations of authenticity/privacy settings." + docs["vendorurl"] = "http://www.ibm.com" + show_docs(options, docs) + + # Operate the fencing device + result = fence_action(FencingSnmp(options), options, set_power_status, get_power_status, get_outlets_status) + + sys.exit(result) +if __name__ == "__main__": + main()

1 0

fence-agents: master - fence agents: Some agents do not support action=metadata on STDIN
by Marek Grác 23 Jul '12

23 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=efcf34f5… Commit: efcf34f5156385c397509f9fe40ae08ca74a96a0 Parent: 723b70ae3f2930e26a7646a4eac881ae1f8e8f10 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Thu Jul 12 18:30:48 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Thu Jul 12 18:30:48 2012 +0200 fence agents: Some agents do not support action=metadata on STDIN Resolves rhbz#837174 --- fence/agents/baytech/fence_baytech.pl | 9 +++++---- fence/agents/brocade/fence_brocade.pl | 5 +++++ fence/agents/bullpap/fence_bullpap.pl | 7 ++++++- fence/agents/cpint/fence_cpint.pl | 10 ++++++++++ fence/agents/drac/fence_drac.pl | 5 +++++ fence/agents/egenera/fence_egenera.pl | 9 +++++---- fence/agents/mcdata/fence_mcdata.pl | 7 ++++++- fence/agents/rackswitch/do_rack.c | 9 +++++++++ fence/agents/xcat/fence_xcat.pl | 5 +++++ fence/agents/zvm/fence_zvm.pl | 10 ++++++++++ 10 files changed, 66 insertions(+), 10 deletions(-) diff --git a/fence/agents/baytech/fence_baytech.pl b/fence/agents/baytech/fence_baytech.pl index c978ec5..e218b78 100644 --- a/fence/agents/baytech/fence_baytech.pl +++ b/fence/agents/baytech/fence_baytech.pl @@ -166,14 +166,15 @@ sub get_options fail_usage "Unkown parameter." if (@ARGV > 0); - if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { - print_metadata(); - exit 0; - } } else { get_options_stdin(); } + if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "failed: must specify hostname" unless defined $opt_a; $host=$opt_a; $port=23 unless ($opt_a =~ /:/); diff --git a/fence/agents/brocade/fence_brocade.pl b/fence/agents/brocade/fence_brocade.pl index e3d1612..e3f7290 100644 --- a/fence/agents/brocade/fence_brocade.pl +++ b/fence/agents/brocade/fence_brocade.pl @@ -162,6 +162,11 @@ if (@ARGV > 0) { } else { get_options_stdin(); + if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "failed: no IP address" unless defined $opt_a; fail "failed: no plug number" unless defined $opt_n; fail "failed: no login name" unless defined $opt_l; diff --git a/fence/agents/bullpap/fence_bullpap.pl b/fence/agents/bullpap/fence_bullpap.pl index 3309c18..e31ab6d 100644 --- a/fence/agents/bullpap/fence_bullpap.pl +++ b/fence/agents/bullpap/fence_bullpap.pl @@ -179,7 +179,7 @@ sub get_options_stdin { $login = $val; } - elsif ($name eq "option" ) + elsif (($name eq "option" ) || ($name eq "action")) { $action = $val; } @@ -312,6 +312,11 @@ else { get_options_stdin(); + if ((defined $action) && ($action =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "failed: no IP address" unless defined $host; fail "failed: no domain" unless defined $domain; fail "failed: no login name" unless defined $login; diff --git a/fence/agents/cpint/fence_cpint.pl b/fence/agents/cpint/fence_cpint.pl index d3c016a..eb2ad3a 100644 --- a/fence/agents/cpint/fence_cpint.pl +++ b/fence/agents/cpint/fence_cpint.pl @@ -142,6 +142,11 @@ sub get_options_stdin $opt_u = $val; } + elsif ($name eq "action" ) + { + $opt_o = $val; + } + else { print stderr "parse error: unknown option \"$opt\"\n"; @@ -166,6 +171,11 @@ if (@ARGV > 0){ } else { get_options_stdin(); + if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "no userid" unless defined $opt_u; } diff --git a/fence/agents/drac/fence_drac.pl b/fence/agents/drac/fence_drac.pl index 825391f..b26fbcb 100644 --- a/fence/agents/drac/fence_drac.pl +++ b/fence/agents/drac/fence_drac.pl @@ -717,6 +717,11 @@ if (@ARGV > 0) { } else { get_options_stdin(); + if ((defined $action) && ($action =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "failed: no IP address" unless defined $address; fail "failed: no login name" unless defined $login; diff --git a/fence/agents/egenera/fence_egenera.pl b/fence/agents/egenera/fence_egenera.pl index 30eaca1..308338b 100644 --- a/fence/agents/egenera/fence_egenera.pl +++ b/fence/agents/egenera/fence_egenera.pl @@ -139,10 +139,6 @@ if (@ARGV > 0) fail_usage "Unkown parameter." if (@ARGV > 0); - if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { - print_metadata(); - exit 0; - } $cserv = $opt_c if defined $opt_c; $lpan = $opt_l if defined $opt_l; @@ -156,6 +152,11 @@ else get_options_stdin(); } +if (((defined $opt_o) && ($opt_o =~ /metadata/i)) || ((defined $action) && ($action =~ /metadata/i))) { + print_metadata(); + exit 0; +} + $action = "reboot" unless defined $action; $user = "root" unless defined $user; diff --git a/fence/agents/mcdata/fence_mcdata.pl b/fence/agents/mcdata/fence_mcdata.pl index a469a1c..5bc9dcf 100644 --- a/fence/agents/mcdata/fence_mcdata.pl +++ b/fence/agents/mcdata/fence_mcdata.pl @@ -176,7 +176,7 @@ sub get_options_stdin { $opt_l = $val; } - elsif ($name eq "option" ) + elsif (($name eq "option" ) || ($name eq "action")) { $opt_o = $val; } @@ -235,6 +235,11 @@ if (@ARGV > 0) { } else { get_options_stdin(); + if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "failed: no IP address" unless defined $opt_a; fail "failed: no plug number" unless defined $opt_n; fail "failed: no login name" unless defined $opt_l; diff --git a/fence/agents/rackswitch/do_rack.c b/fence/agents/rackswitch/do_rack.c index 67d24a2..d8050bf 100644 --- a/fence/agents/rackswitch/do_rack.c +++ b/fence/agents/rackswitch/do_rack.c @@ -301,6 +301,15 @@ static void get_options(int argc, char **argv) strcpy(name, pname); if (!strcmp(arg, "ipaddr")) strcpy(ipaddr, value); + + if (!strcmp(arg, "action")) + if (strncasecmp(value, "metadata", 254) == 0) { + print_metadata(); + exit(DID_SUCCESS); + } else { + fprintf(stderr, "Only 'metadata' option is aviable for this fence agent\n"); + exit(DID_FAILURE); + } if (!strcmp(arg, "portnumber")) strcpy(portnumber, value); diff --git a/fence/agents/xcat/fence_xcat.pl b/fence/agents/xcat/fence_xcat.pl index f984ec1..3cd16aa 100644 --- a/fence/agents/xcat/fence_xcat.pl +++ b/fence/agents/xcat/fence_xcat.pl @@ -184,6 +184,11 @@ if (@ARGV > 0) { } else { get_options_stdin(); + if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "failed: no plug number" unless defined $opt_n; $opt_o=lc($opt_o); fail "failed: unrecognised action: $opt_o" diff --git a/fence/agents/zvm/fence_zvm.pl b/fence/agents/zvm/fence_zvm.pl index e6386e4..04a29e2 100644 --- a/fence/agents/zvm/fence_zvm.pl +++ b/fence/agents/zvm/fence_zvm.pl @@ -299,6 +299,11 @@ sub get_options_stdin { $opt_r = $val; } + + elsif ($name eq "action") + { + $opt_o = $val; + } # FIXME -- depreicated residue of old fencing system elsif ($name eq "name" ) { } @@ -352,6 +357,11 @@ if (@ARGV > 0){ } else { get_options_stdin(); + if ((defined $opt_o) && ($opt_o =~ /metadata/i)) { + print_metadata(); + exit 0; + } + fail "no IP address" unless defined $opt_a; fail "no userid" unless defined $opt_u;

1 0

cluster: RHEL59 - When the filesystem /etc lives on is completely full, umount will exit with exit status 16 if the umount syscall succeeded but it was unable to write a new mtab file because the disk is full. umount won't exit with status 16 under any other circumstances.
by Ryan McCabe 16 Jul '12

16 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=3290dff9b8899… Commit: 3290dff9b88992e913b28dd57c6b64a3c2b75c8a Parent: 13005d1ff5246a332bfc6795ad443b0148cd11b2 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Mon Jul 16 11:21:43 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Mon Jul 16 11:21:43 2012 -0400 When the filesystem /etc lives on is completely full, umount will exit with exit status 16 if the umount syscall succeeded but it was unable to write a new mtab file because the disk is full. umount won't exit with status 16 under any other circumstances. This patch changes the fs.sh, clusterfs.sh, and netfs.sh resource agents to check treat both exit status 0 and exit status 16 as success. Resolves: rhbz#819595 Acked-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/resources/clusterfs.sh | 5 ++++- rgmanager/src/resources/fs.sh | 5 ++++- rgmanager/src/resources/netfs.sh | 5 ++++- 3 files changed, 12 insertions(+), 3 deletions(-) diff --git a/rgmanager/src/resources/clusterfs.sh b/rgmanager/src/resources/clusterfs.sh index 49eb724..ff28e3d 100755 --- a/rgmanager/src/resources/clusterfs.sh +++ b/rgmanager/src/resources/clusterfs.sh @@ -792,8 +792,11 @@ stop: Could not match $OCF_RESKEY_device with a real device" sync; sync; sync ocf_log info "unmounting $dev ($mp)" + # umount will exit with status 16 iff the umount(2) + # succeeded, but /etc/mtab could not be written. umount $mp - if [ $? -eq 0 ]; then + retval=$? + if [ $retval -eq 0 -o $retval -eq 16 ]; then umount_failed= done=$YES continue diff --git a/rgmanager/src/resources/fs.sh b/rgmanager/src/resources/fs.sh index a98cddc..9dec8d1 100755 --- a/rgmanager/src/resources/fs.sh +++ b/rgmanager/src/resources/fs.sh @@ -1102,8 +1102,11 @@ stop: Could not match $OCF_RESKEY_device with a real device" fi ocf_log info "unmounting $mp" + # umount will exit with status 16 iff the umount(2) + # succeeded, but /etc/mtab could not be written. umount $mp - if [ $? -eq 0 ]; then + retval=$? + if [ $retval -eq 0 -o $retval -eq 16 ]; then umount_failed= done=$YES continue diff --git a/rgmanager/src/resources/netfs.sh b/rgmanager/src/resources/netfs.sh index 837a4c4..6089a0d 100755 --- a/rgmanager/src/resources/netfs.sh +++ b/rgmanager/src/resources/netfs.sh @@ -559,8 +559,11 @@ stopNFSFilesystem() { sync; sync; sync ocf_log info "unmounting $mp" + # umount will exit with status 16 iff the umount(2) + # succeeded, but /etc/mtab could not be written. umount $umount_flag $mp - if [ $? -eq 0 ]; then + retval=$? + if [ $retval -eq 0 -o $retval -eq 16 ]; then umount_failed= done=$YES continue

1 0

cluster: RHEL59 - rgmanager: Fix return code when relocation fails and svc is running on original node
by Ryan McCabe 16 Jul '12

16 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=13005d1ff5246… Commit: 13005d1ff5246a332bfc6795ad443b0148cd11b2 Parent: 467d015c53e5e2d00025708cda95b3df41ebda1f Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Wed Jul 11 09:11:33 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Mon Jul 16 11:09:20 2012 -0400 rgmanager: Fix return code when relocation fails and svc is running on original node Return RG_ERELO (Relocation failure; service running on original node) when relocation failed and the service was restarted successfully on the original node. This is the behavior that's currently in the STABLE32 and RHEL6 branches. Acked-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/daemons/rg_state.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/rgmanager/src/daemons/rg_state.c b/rgmanager/src/daemons/rg_state.c index 4357d21..9000f1b 100644 --- a/rgmanager/src/daemons/rg_state.c +++ b/rgmanager/src/daemons/rg_state.c @@ -2061,7 +2061,7 @@ exhausted: svcName); if (svc_start(svcName, RG_START_RECOVER) == 0) { *new_owner = me; - return 0; + return RG_ERELO; } }

1 0

cluster: RHEL59 - rgmanager: Fix for services stuck in recovery
by Ryan McCabe 16 Jul '12

16 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=467d015c53e5e… Commit: 467d015c53e5e2d00025708cda95b3df41ebda1f Parent: d88584bd640700c51692198d2f6aeda0e773165c Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Mon Jul 16 10:57:28 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Mon Jul 16 11:05:26 2012 -0400 rgmanager: Fix for services stuck in recovery Patch from John Ruemker <jruemker(a)redhat.com>: "When starting rgmanager throughout the cluster around the same time, multiple nodes may end up acting as the "root" for a particular service. If that service happens to fail on startup, you can end up with each of those nodes sending remote-start requests around the cluster. Eventually the service will get stuck in a recovering state, and cannot be modified in any way with clusvcadm. The only remedy we've found is to kill rgmanager and start it back up. Acked-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/daemons/groups.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/rgmanager/src/daemons/groups.c b/rgmanager/src/daemons/groups.c index 20ed2e1..bd406c8 100644 --- a/rgmanager/src/daemons/groups.c +++ b/rgmanager/src/daemons/groups.c @@ -747,7 +747,8 @@ eval_groups(int local, uint32_t nodeid, int nodeStatus) (svcStatus.rs_state == RG_STATE_STARTED || svcStatus.rs_state == RG_STATE_RECOVER || svcStatus.rs_state == RG_STATE_STARTING || - svcStatus.rs_state == RG_STATE_STOPPING )) { + svcStatus.rs_state == RG_STATE_STOPPING || + svcStatus.rs_state == RG_STATE_ERROR)) { clulog(LOG_DEBUG, "Marking %s on down member %d as stopped", @@ -789,7 +790,8 @@ eval_groups(int local, uint32_t nodeid, int nodeStatus) /* Disabled/failed/in recovery? Do nothing */ if ((svcStatus.rs_state == RG_STATE_DISABLED) || (svcStatus.rs_state == RG_STATE_FAILED) || - (svcStatus.rs_state == RG_STATE_RECOVER)) { + (svcStatus.rs_state == RG_STATE_RECOVER) || + (svcStatus.rs_state == RG_STATE_ERROR)) { continue; }

1 0

cluster: STABLE32 - rgmanager: Fix for services stuck in recovery
by Ryan McCabe 16 Jul '12

16 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=03e2215bd277f… Commit: 03e2215bd277fd79b8a6ee70a49de711e0f343ad Parent: e080a6249f25a3bc18a6c9028c9da06bf4ef14c6 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Mon Jul 16 10:57:28 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Mon Jul 16 10:57:28 2012 -0400 rgmanager: Fix for services stuck in recovery Patch from John Ruemker <jruemker(a)redhat.com>: "When starting rgmanager throughout the cluster around the same time, multiple nodes may end up acting as the "root" for a particular service. If that service happens to fail on startup, you can end up with each of those nodes sending remote-start requests around the cluster. Eventually the service will get stuck in a recovering state, and cannot be modified in any way with clusvcadm. The only remedy we've found is to kill rgmanager and start it back up. Acked-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/daemons/groups.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/rgmanager/src/daemons/groups.c b/rgmanager/src/daemons/groups.c index 4a72cb7..4537cc8 100644 --- a/rgmanager/src/daemons/groups.c +++ b/rgmanager/src/daemons/groups.c @@ -731,7 +731,8 @@ eval_groups(int local, uint32_t nodeid, int nodeStatus) (svcStatus.rs_state == RG_STATE_STARTED || svcStatus.rs_state == RG_STATE_RECOVER || svcStatus.rs_state == RG_STATE_STARTING || - svcStatus.rs_state == RG_STATE_STOPPING )) { + svcStatus.rs_state == RG_STATE_STOPPING || + svcStatus.rs_state == RG_STATE_ERROR)) { logt_print(LOG_DEBUG, "Marking %s on down member %d as stopped", @@ -773,7 +774,8 @@ eval_groups(int local, uint32_t nodeid, int nodeStatus) /* Disabled/failed/in recovery? Do nothing */ if ((svcStatus.rs_state == RG_STATE_DISABLED) || (svcStatus.rs_state == RG_STATE_FAILED) || - (svcStatus.rs_state == RG_STATE_RECOVER)) { + (svcStatus.rs_state == RG_STATE_RECOVER) || + (svcStatus.rs_state == RG_STATE_ERROR)) { continue; }

1 0

cluster: RHEL59 - fsck.gfs2: Fix buffer overflow in get_lockproto_table
by Andrew Price 16 Jul '12

16 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=d88584bd64070… Commit: d88584bd640700c51692198d2f6aeda0e773165c Parent: 5a1d50a8308bb5b40c6a8d990f628bdaa9c20a59 Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Sat Jul 7 22:03:24 2012 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Mon Jul 16 14:18:08 2012 +0100 fsck.gfs2: Fix buffer overflow in get_lockproto_table Coverity discovered a buffer overflow in this function where an overly long cluster name in cluster.conf could cause a crash while repairing the superblock. This patch fixes the bug by making sure the lock table is composed sensibly, limiting the fsname to 16 chars as documented, and only allowing the cluster name (which doesn't seem to have a documented max size) to use the remaining space in the locktable name string. Resolves: rhbz#838910 Signed-off-by: Andrew Price <anprice(a)redhat.com> --- gfs2/fsck/initialize.c | 42 ++++++++++++++++++++++-------------------- 1 files changed, 22 insertions(+), 20 deletions(-) diff --git a/gfs2/fsck/initialize.c b/gfs2/fsck/initialize.c index bd364e4..a26ed84 100644 --- a/gfs2/fsck/initialize.c +++ b/gfs2/fsck/initialize.c @@ -551,12 +551,13 @@ static int init_system_inodes(struct gfs2_sbd *sdp) static int get_lockproto_table(struct gfs2_sbd *sdp) { FILE *fp; - char line[PATH_MAX], *p, *p2; - char fsname[PATH_MAX]; + char line[PATH_MAX]; + char *cluname, *end; + const char *fsname, *cfgfile = "/etc/cluster/cluster.conf"; memset(sdp->lockproto, 0, sizeof(sdp->lockproto)); memset(sdp->locktable, 0, sizeof(sdp->locktable)); - fp = fopen("/etc/cluster/cluster.conf", "rt"); + fp = fopen(cfgfile, "rt"); if (!fp) { /* no cluster.conf; must be a stand-alone file system */ strcpy(sdp->lockproto, "lock_nolock"); @@ -569,28 +570,29 @@ static int get_lockproto_table(struct gfs2_sbd *sdp) log_warn(_("Lock protocol assumed to be: " GFS2_DEFAULT_LOCKPROTO "\n")); strcpy(sdp->lockproto, GFS2_DEFAULT_LOCKPROTO); + while (fgets(line, sizeof(line) - 1, fp)) { - p = strstr(line,"<cluster name="); - if (p) { - p += 15; - p2 = strchr(p,'"'); - strncpy(sdp->locktable, p, p2 - p); + cluname = strstr(line,"<cluster name="); + if (cluname) { + cluname += 15; + end = strchr(cluname,'"'); + if (end) + *end = '\0'; break; } } - if (sdp->locktable[0] == '\0') { - log_err(_("Error: Unable to determine cluster name from " - "/etc/cluster.conf\n")); + if (cluname == NULL || end == NULL || end - cluname < 1) { + log_err(_("Error: Unable to determine cluster name from %s\n"), + cfgfile); } else { - memset(fsname, 0, sizeof(fsname)); - p = strrchr(opts.device, '/'); - if (p) { - p++; - strncpy(fsname, p, sizeof(fsname)); - } else - strcpy(fsname, "repaired"); - strcat(sdp->locktable, ":"); - strcat(sdp->locktable, fsname); + fsname = strrchr(opts.device, '/'); + if (fsname) + fsname++; + else + fsname = "repaired"; + snprintf(sdp->locktable, sizeof(sdp->locktable), "%.*s:%.16s", + (int)(sizeof(sdp->locktable) - strlen(fsname) - 2), + cluname, fsname); log_warn(_("Lock table determined to be: %s\n"), sdp->locktable); }

1 0

gfs2-utils: master - fsck.gfs2: Fix buffer overflow in get_lockproto_table
by Andrew Price 12 Jul '12

12 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=gfs2-utils.git;a=commitdiff;h=150b51604f… Commit: 150b51604f75c0de65cf3f837bc7b8ac30aa6d8b Parent: 04bb1157cd4967c3d935a3c77c9ba0f176dcff43 Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Sat Jul 7 22:03:24 2012 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Sat Jul 7 22:03:24 2012 +0100 fsck.gfs2: Fix buffer overflow in get_lockproto_table Coverity discovered a buffer overflow in this function where an overly long cluster name in cluster.conf could cause a crash while repairing the superblock. This patch fixes the bug by making sure the lock table is composed sensibly, limiting the fsname to 16 chars as documented, and only allowing the cluster name (which doesn't seem to have a documented max size) to use the remaining space in the locktable name string. Signed-off-by: Andrew Price <anprice(a)redhat.com> --- gfs2/fsck/initialize.c | 42 ++++++++++++++++++++++-------------------- 1 files changed, 22 insertions(+), 20 deletions(-) diff --git a/gfs2/fsck/initialize.c b/gfs2/fsck/initialize.c index f07e0b2..e56161e 100644 --- a/gfs2/fsck/initialize.c +++ b/gfs2/fsck/initialize.c @@ -753,12 +753,13 @@ static int init_system_inodes(struct gfs2_sbd *sdp) static int get_lockproto_table(struct gfs2_sbd *sdp) { FILE *fp; - char line[PATH_MAX], *p, *p2; - char fsname[PATH_MAX]; + char line[PATH_MAX]; + char *cluname, *end; + const char *fsname, *cfgfile = "/etc/cluster/cluster.conf"; memset(sdp->lockproto, 0, sizeof(sdp->lockproto)); memset(sdp->locktable, 0, sizeof(sdp->locktable)); - fp = fopen("/etc/cluster/cluster.conf", "rt"); + fp = fopen(cfgfile, "rt"); if (!fp) { /* no cluster.conf; must be a stand-alone file system */ strcpy(sdp->lockproto, "lock_nolock"); @@ -771,28 +772,29 @@ static int get_lockproto_table(struct gfs2_sbd *sdp) log_warn(_("Lock protocol assumed to be: " GFS2_DEFAULT_LOCKPROTO "\n")); strcpy(sdp->lockproto, GFS2_DEFAULT_LOCKPROTO); + while (fgets(line, sizeof(line) - 1, fp)) { - p = strstr(line,"<cluster name="); - if (p) { - p += 15; - p2 = strchr(p,'"'); - strncpy(sdp->locktable, p, p2 - p); + cluname = strstr(line,"<cluster name="); + if (cluname) { + cluname += 15; + end = strchr(cluname,'"'); + if (end) + *end = '\0'; break; } } - if (sdp->locktable[0] == '\0') { - log_err(_("Error: Unable to determine cluster name from " - "/etc/cluster.conf\n")); + if (cluname == NULL || end == NULL || end - cluname < 1) { + log_err(_("Error: Unable to determine cluster name from %s\n"), + cfgfile); } else { - memset(fsname, 0, sizeof(fsname)); - p = strrchr(opts.device, '/'); - if (p) { - p++; - strncpy(fsname, p, sizeof(fsname)); - } else - strcpy(fsname, "repaired"); - strcat(sdp->locktable, ":"); - strcat(sdp->locktable, fsname); + fsname = strrchr(opts.device, '/'); + if (fsname) + fsname++; + else + fsname = "repaired"; + snprintf(sdp->locktable, sizeof(sdp->locktable), "%.*s:%.16s", + (int)(sizeof(sdp->locktable) - strlen(fsname) - 2), + cluname, fsname); log_warn(_("Lock table determined to be: %s\n"), sdp->locktable); }

1 0

gfs2-utils: master - mkfs.gfs2: Avoid a rename race when checking file contents
by Andrew Price 12 Jul '12

12 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=gfs2-utils.git;a=commitdiff;h=04bb1157cd… Commit: 04bb1157cd4967c3d935a3c77c9ba0f176dcff43 Parent: aaa9d0c9f75936ef8cc19d8cfc4f48f8b4bfdd59 Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Fri Jul 6 12:50:12 2012 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Fri Jul 6 12:50:12 2012 +0100 mkfs.gfs2: Avoid a rename race when checking file contents Currently there is a slight chance that mkfs.gfs2 could report the incorrect contents of the target if another file is renamed into place. This is because we pass the target name to the file command and file opens it again. This patch tries to avoid that condition by using the /proc/$pid/fd/$device_fd symlink, which changes when the open file is renamed, instead. Signed-off-by: Andrew Price <anprice(a)redhat.com> --- gfs2/mkfs/main_mkfs.c | 11 +++++++++-- 1 files changed, 9 insertions(+), 2 deletions(-) diff --git a/gfs2/mkfs/main_mkfs.c b/gfs2/mkfs/main_mkfs.c index 957b144..2d529d7 100644 --- a/gfs2/mkfs/main_mkfs.c +++ b/gfs2/mkfs/main_mkfs.c @@ -426,7 +426,7 @@ static void check_dev_content(const char *devname) char content[1024] = { 0, }; char * args[] = { (char *)"/usr/bin/file", - (char *)"-bs", + (char *)"-bsL", (char *)devname, NULL }; int p[2] = {-1, -1}; @@ -557,6 +557,7 @@ void main_mkfs(int argc, char *argv[]) int rgsize_specified = 0; unsigned char uuid[16]; char *absname = NULL; + char *fdpath = NULL; int islnk = 0; memset(sdp, 0, sizeof(struct gfs2_sbd)); @@ -587,13 +588,19 @@ void main_mkfs(int argc, char *argv[]) exit(EXIT_FAILURE); } + if (asprintf(&fdpath, "/proc/%d/fd/%d", getpid(), sdp->device_fd) < 0) { + perror(_("Failed to build string")); + exit(EXIT_FAILURE); + } + if (!sdp->override) { islnk = is_symlink(sdp->device_name, &absname); printf(_("This will destroy any data on %s.\n"), islnk ? absname : sdp->device_name); - check_dev_content(islnk ? absname : sdp->device_name); free(absname); + check_dev_content(fdpath); are_you_sure(); } + free(fdpath); if (sdp->bsize == -1) { if (S_ISREG(sdp->dinfo.stat.st_mode))

1 0

cluster: STABLE32 - cman: fix a couple of comments
by Fabio M. Di Nitto 12 Jul '12

12 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=e080a6249f25a… Commit: e080a6249f25a3bc18a6c9028c9da06bf4ef14c6 Parent: 88ab6a7b1e4da3802f64addc3d6f234e5cf82760 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Thu Jul 12 13:00:27 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Thu Jul 12 13:00:27 2012 +0200 cman: fix a couple of comments Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/daemon/cman-preconfig.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index 22583fe..08b5ecf 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -1533,7 +1533,7 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const objdb->object_destroy(cluster_parent_handle); /* - * create cluster.cman in the new config if it doesn't exists + * create cluster.cman in the new config if it doesn't exist */ objdb->object_find_create(cluster_parent_handle_new, "cman", strlen("cman"), &find_handle); if (objdb->object_find_next(find_handle, &object_handle)) { @@ -1543,7 +1543,7 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const objdb->object_find_destroy(find_handle); /* - * readd cluster_id/two_node/nodename + * write cluster_id/two_node/nodename */ snprintf(str, sizeof(str) - 1, "%d", cluster_id); objdb->object_key_create_typed(object_handle, "cluster_id",

1 0

cluster: STABLE32 - cman: fix data copy and memory leak when reloading config
by Fabio M. Di Nitto 11 Jul '12

11 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=88ab6a7b1e4da… Commit: 88ab6a7b1e4da3802f64addc3d6f234e5cf82760 Parent: 8975bd6341b2d94c1f89279b1b00d4360da1f5ff Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Wed Jul 11 11:44:49 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Wed Jul 11 11:44:49 2012 +0200 cman: fix data copy and memory leak when reloading config cman.cluster_id,nodename,two_node information were not copied from the old to the new config at reload time. This triggers a problem when <cman is set in cluster.conf and we effectively drop information from objdb (suboptimal). Also fix a possible memory leak when we have reload issues. Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/daemon/cman-preconfig.c | 31 +++++++++++++++++++++++++++++-- 1 files changed, 29 insertions(+), 2 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index 68fec22..22583fe 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -1478,6 +1478,7 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const hdb_handle_t cluster_parent_handle_new; unsigned int config_version = 0, config_version_new = 0; char *config_value = NULL; + char str[255]; /* don't reload if we've been told to run configless */ if (getenv("CMAN_NOCONFIG")) { @@ -1489,16 +1490,16 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const /* find both /cluster entries */ objdb->object_find_create(OBJECT_PARENT_HANDLE, "cluster", strlen("cluster"), &find_handle); objdb->object_find_next(find_handle, &cluster_parent_handle); + objdb->object_find_next(find_handle, &cluster_parent_handle_new); + objdb->object_find_destroy(find_handle); if (!cluster_parent_handle) { snprintf (error_reason, sizeof(error_reason) - 1, "Cannot find old /cluster/ key in configuration\n"); goto err; } - objdb->object_find_next(find_handle, &cluster_parent_handle_new); if (!cluster_parent_handle_new) { snprintf (error_reason, sizeof(error_reason) - 1, "Cannot find new /cluster/ key in configuration\n"); goto err; } - objdb->object_find_destroy(find_handle); if (!objdb->object_key_get(cluster_parent_handle, "config_version", strlen("config_version"), (void *)&config_value, NULL)) { if (config_value) { @@ -1531,6 +1532,32 @@ static int cmanpre_reloadconfig(struct objdb_iface_ver0 *objdb, int flush, const /* destroy the old one */ objdb->object_destroy(cluster_parent_handle); + /* + * create cluster.cman in the new config if it doesn't exists + */ + objdb->object_find_create(cluster_parent_handle_new, "cman", strlen("cman"), &find_handle); + if (objdb->object_find_next(find_handle, &object_handle)) { + objdb->object_create(cluster_parent_handle_new, &object_handle, + "cman", strlen("cman")); + } + objdb->object_find_destroy(find_handle); + + /* + * readd cluster_id/two_node/nodename + */ + snprintf(str, sizeof(str) - 1, "%d", cluster_id); + objdb->object_key_create_typed(object_handle, "cluster_id", + str, strlen(str) + 1, OBJDB_VALUETYPE_STRING); + + if (two_node) { + snprintf(str, sizeof(str) - 1, "%d", 1); + objdb->object_key_create_typed(object_handle, "two_node", + str, strlen(str) + 1, OBJDB_VALUETYPE_STRING); + } + + objdb->object_key_create_typed(object_handle, "nodename", + nodename, strlen(nodename)+1, OBJDB_VALUETYPE_STRING); + /* update the reference to the new config */ cluster_parent_handle = cluster_parent_handle_new;

1 0

cluster: RHEL59 - rgmanager: Don't immediately stop services that have started sucessfully
by Ryan McCabe 10 Jul '12

10 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=5a1d50a8308bb… Commit: 5a1d50a8308bb5b40c6a8d990f628bdaa9c20a59 Parent: 55710722d15be8f2eafdae472086182f88b2a0d5 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Thu Jun 28 15:24:48 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Tue Jul 10 15:56:33 2012 -0400 rgmanager: Don't immediately stop services that have started sucessfully This patch fixes a bug that caused services to be started, then immediately stopped during recovery. This occurred when a service had its recovery policy set to relocate and was in a restricted failover domain. If starting the service failed on other nodes in the failover domain, the service would be restarted on the original node, then immediately stopped, even if it had started successfully. This patch causes rgmanager to leave the service running if it restarted successfully. Resolves: rhbz#789366 Acked-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/daemons/rg_state.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/rgmanager/src/daemons/rg_state.c b/rgmanager/src/daemons/rg_state.c index a8b1e36..4357d21 100644 --- a/rgmanager/src/daemons/rg_state.c +++ b/rgmanager/src/daemons/rg_state.c @@ -2061,7 +2061,7 @@ exhausted: svcName); if (svc_start(svcName, RG_START_RECOVER) == 0) { *new_owner = me; - return RG_EFAIL; + return 0; } }

1 0

cluster: RHEL59 - rgmanager: Don't exit uncleanly when cman asks us to shut down.
by Ryan McCabe 10 Jul '12

10 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=55710722d15be… Commit: 55710722d15be8f2eafdae472086182f88b2a0d5 Parent: df45c13985a5219c04356eea98179935ed9efb14 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Thu Jun 28 15:18:09 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Tue Jul 10 15:55:29 2012 -0400 rgmanager: Don't exit uncleanly when cman asks us to shut down. Original patch from Lon rediffed to apply to the current tree: "Previous to this, rgmanager would uncleanly exit if you issued a 'service cman stop'. This patch makes it uncleanly exit if 'cman_tool leave force' or a corosync/openais crash occurs, but in a simple cman_tool leave, rgmanager will no longer exit uncleanly." Without this patch, issuing 'service cman stop' when rgmanager is running will make it impossible to stop the cman service because rgmanager will have exited without releasing its dlm lockspace. This patch causes rgmanager to refuse shutdown when requested by cman. Users must stop the rgmanager service before stopping the cman service. Acked-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/clulib/msg_cluster.c | 7 ++++++- 1 files changed, 6 insertions(+), 1 deletions(-) diff --git a/rgmanager/src/clulib/msg_cluster.c b/rgmanager/src/clulib/msg_cluster.c index 8dc22d0..e864853 100644 --- a/rgmanager/src/clulib/msg_cluster.c +++ b/rgmanager/src/clulib/msg_cluster.c @@ -211,7 +211,7 @@ poll_cluster_messages(int timeout) if (cman_dispatch(ch, 0) < 0) { process_cman_event(ch, NULL, - CMAN_REASON_TRY_SHUTDOWN, 0); + CMAN_REASON_TRY_SHUTDOWN, 1); } ret = 0; } @@ -987,6 +987,11 @@ process_cman_event(cman_handle_t handle, void *private, int reason, int arg) printf("EVENT: %p %p %d %d\n", handle, private, reason, arg); #endif + if (reason == CMAN_REASON_TRY_SHUTDOWN && !arg) { + cman_replyto_shutdown(handle, 0); + return; + } + /* Allocate queue node */ while ((node = malloc(sizeof(*node))) == NULL) { sleep(1);

1 0

cluster: RHEL59 - rgmanager: Fix orainstance.sh error checking
by Ryan McCabe 10 Jul '12

10 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=df45c13985a52… Commit: df45c13985a5219c04356eea98179935ed9efb14 Parent: 49c6289cd6c394f6b18d7f61f0177ed886643f14 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Thu Jun 28 11:19:16 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Tue Jul 10 15:54:07 2012 -0400 rgmanager: Fix orainstance.sh error checking Pull in the fixed error checking that was added to oracledb.sh as a fix for rhbz#471066. Resolves rhbz#723819 Acked-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/resources/orainstance.sh | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/rgmanager/src/resources/orainstance.sh b/rgmanager/src/resources/orainstance.sh index 6f2ff15..a9f690d 100755 --- a/rgmanager/src/resources/orainstance.sh +++ b/rgmanager/src/resources/orainstance.sh @@ -105,7 +105,7 @@ start_db() { # If we see: # ORA-.....: failure, we failed - grep -q "failure" $logfile + grep -q "^ORA-" $logfile rv=$? rm -f $logfile @@ -155,7 +155,7 @@ stop_db() { return 1 fi - grep -q failure $logfile + grep -q "^ORA-" $logfile rv=$? rm -f $logfile

1 0

cluster: RHEL59 - rgmanager: Fix a possible NULL pointer dereference
by Ryan McCabe 10 Jul '12

10 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=49c6289cd6c39… Commit: 49c6289cd6c394f6b18d7f61f0177ed886643f14 Parent: 4a3415a2056b8b09ab272edefc0de8950b84f393 Author: Ryan McCabe <rmccabe(a)redhat.com> AuthorDate: Thu Jun 28 15:50:37 2012 -0400 Committer: Ryan McCabe <rmccabe(a)redhat.com> CommitterDate: Tue Jul 10 15:52:14 2012 -0400 rgmanager: Fix a possible NULL pointer dereference Fix a NULL pointer dereference that could happen when cman_get_node_count() returns 0 with errno set to EINTR. Possibly resolves rhbz#820632 Acked-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Signed-off-by: Ryan McCabe <rmccabe(a)redhat.com> --- rgmanager/src/clulib/members.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/rgmanager/src/clulib/members.c b/rgmanager/src/clulib/members.c index f705297..72f4529 100644 --- a/rgmanager/src/clulib/members.c +++ b/rgmanager/src/clulib/members.c @@ -367,8 +367,10 @@ get_member_list(cman_handle_t h) do { ++tries; - if (nodes) + if (nodes) { free(nodes); + nodes = NULL; + } c = cman_get_node_count(h); if (c <= 0) {

1 0

gfs2-utils: master - gfs2-utils: Remove unused exported functions
by Andrew Price 04 Jul '12

04 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=gfs2-utils.git;a=commitdiff;h=aaa9d0c9f7… Commit: aaa9d0c9f75936ef8cc19d8cfc4f48f8b4bfdd59 Parent: 3b748462ec98cb5909b3ad9f3b639dbca38142fd Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Wed Jul 4 14:58:18 2012 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Wed Jul 4 14:58:18 2012 +0100 gfs2-utils: Remove unused exported functions These symbols are now unused within the gfs2-utils tree: $ findunused.sh | grep -v '^group/' gfs2/libgfs2/libgfs2.h:712 find_debugfs_mount gfs2/libgfs2/misc.c:314 find_debugfs_mount gfs2/libgfs2/libgfs2.h:395 gfs2_bitcount gfs2/libgfs2/fs_bits.c:109 gfs2_bitcount gfs2/libgfs2/libgfs2.h:835 gfs2_dirent_print gfs2/libgfs2/ondisk.c:418 gfs2_dirent_print gfs2/libgfs2/libgfs2.h:810 gfs2_ea_header_out gfs2/libgfs2/ondisk.c:481 gfs2_ea_header_out gfs2/libgfs2/fs_ops.c:1371 init_gfs_dinode gfs2/libgfs2/libgfs2.h:451 init_gfs_dinode gfs2/libgfs2/libgfs2.h:713 mp2fsname gfs2/libgfs2/misc.c:365 mp2fsname gfs2/libgfs2/libgfs2.h:705 get_sysfs gfs2/libgfs2/misc.c:273 get_sysfs So it's safe to remove them. The idea is to simplify libgfs2 for now but if these are required in future they can be brought back. Signed-off-by: Andrew Price <anprice(a)redhat.com> --- gfs2/libgfs2/fs_bits.c | 32 ------------ gfs2/libgfs2/fs_ops.c | 8 --- gfs2/libgfs2/libgfs2.h | 12 ----- gfs2/libgfs2/misc.c | 124 ------------------------------------------------ gfs2/libgfs2/ondisk.c | 28 ----------- 5 files changed, 0 insertions(+), 204 deletions(-) diff --git a/gfs2/libgfs2/fs_bits.c b/gfs2/libgfs2/fs_bits.c index d3ac048..5eeb920 100644 --- a/gfs2/libgfs2/fs_bits.c +++ b/gfs2/libgfs2/fs_bits.c @@ -98,38 +98,6 @@ unsigned long gfs2_bitfit(const unsigned char *buf, const unsigned int len, return (((const unsigned char *)ptr - buf) * GFS2_NBBY) + bit; } -/** - * fs_bitcount - count the number of bits in a certain state - * @buffer: the buffer that holds the bitmaps - * @buflen: the length (in bytes) of the buffer - * @state: the state of the block we're looking for - * - * Returns: The number of bits - */ -uint32_t gfs2_bitcount(unsigned char *buffer, unsigned int buflen, - unsigned char state) -{ - unsigned char *byte, *end; - unsigned int bit; - uint32_t count = 0; - - byte = buffer; - bit = 0; - end = buffer + buflen; - - while (byte < end){ - if (((*byte >> bit) & GFS2_BIT_MASK) == state) - count++; - - bit += GFS2_BIT_SIZE; - if (bit >= 8){ - bit = 0; - byte++; - } - } - return count; -} - /* * check_range - check if blkno is within FS limits * @sdp: super block diff --git a/gfs2/libgfs2/fs_ops.c b/gfs2/libgfs2/fs_ops.c index 470916c..9409332 100644 --- a/gfs2/libgfs2/fs_ops.c +++ b/gfs2/libgfs2/fs_ops.c @@ -1368,14 +1368,6 @@ struct gfs2_buffer_head *init_dinode(struct gfs2_sbd *sdp, return __init_dinode(sdp, inum, mode, flags, parent, 0); } -struct gfs2_buffer_head *init_gfs_dinode(struct gfs2_sbd *sdp, - struct gfs2_inum *inum, - unsigned int mode, uint32_t flags, - struct gfs2_inum *parent) -{ - return __init_dinode(sdp, inum, mode, flags, parent, 1); -} - static struct gfs2_inode *__createi(struct gfs2_inode *dip, const char *filename, unsigned int mode, uint32_t flags, int if_gfs1) diff --git a/gfs2/libgfs2/libgfs2.h b/gfs2/libgfs2/libgfs2.h index 2fc47de..09f0b1d 100644 --- a/gfs2/libgfs2/libgfs2.h +++ b/gfs2/libgfs2/libgfs2.h @@ -392,8 +392,6 @@ extern void fix_device_geometry(struct gfs2_sbd *sdp); #define BFITNOENT (0xFFFFFFFF) /* functions with blk #'s that are buffer relative */ -extern uint32_t gfs2_bitcount(unsigned char *buffer, unsigned int buflen, - unsigned char state); extern unsigned long gfs2_bitfit(const unsigned char *buffer, const unsigned int buflen, unsigned long goal, unsigned char old_state); @@ -448,11 +446,6 @@ extern struct gfs2_buffer_head *init_dinode(struct gfs2_sbd *sdp, struct gfs2_inum *inum, unsigned int mode, uint32_t flags, struct gfs2_inum *parent); -extern struct gfs2_buffer_head *init_gfs_dinode(struct gfs2_sbd *sdp, - struct gfs2_inum *inum, - unsigned int mode, - uint32_t flags, - struct gfs2_inum *parent); extern struct gfs2_inode *createi(struct gfs2_inode *dip, const char *filename, unsigned int mode, uint32_t flags); extern struct gfs2_inode *gfs_createi(struct gfs2_inode *dip, @@ -709,9 +702,6 @@ extern int dir_exists(const char *dir); extern int check_for_gfs2(struct gfs2_sbd *sdp); extern int mount_gfs2_meta(struct gfs2_sbd *sdp); extern void cleanup_metafs(struct gfs2_sbd *sdp); -extern char *find_debugfs_mount(void); -extern char *mp2fsname(char *mp); -extern char *get_sysfs(const char *fsname, const char *filename); extern int set_sysfs(const char *fsname, const char *filename, const char *val); extern int is_fsname(char *name); extern void get_random_bytes(void *buf, int nbytes); @@ -807,7 +797,6 @@ extern void gfs2_dirent_out(struct gfs2_dirent *de, char *buf); extern void gfs2_leaf_in(struct gfs2_leaf *lf, struct gfs2_buffer_head *bh); extern void gfs2_leaf_out(struct gfs2_leaf *lf, struct gfs2_buffer_head *bh); extern void gfs2_ea_header_in(struct gfs2_ea_header *ea, char *buf); -extern void gfs2_ea_header_out(struct gfs2_ea_header *ea, char *buf); extern void gfs2_log_header_in(struct gfs2_log_header *lh, struct gfs2_buffer_head *bh); extern void gfs2_log_header_out(struct gfs2_log_header *lh, @@ -832,7 +821,6 @@ extern void gfs2_rindex_print(struct gfs2_rindex *ri); extern void gfs2_rgrp_print(struct gfs2_rgrp *rg); extern void gfs2_quota_print(struct gfs2_quota *qu); extern void gfs2_dinode_print(struct gfs2_dinode *di); -extern void gfs2_dirent_print(struct gfs2_dirent *de, char *name); extern void gfs2_leaf_print(struct gfs2_leaf *lf); extern void gfs2_ea_header_print(struct gfs2_ea_header *ea, char *name); extern void gfs2_log_header_print(struct gfs2_log_header *lh); diff --git a/gfs2/libgfs2/misc.c b/gfs2/libgfs2/misc.c index 13ef972..a68da4a 100644 --- a/gfs2/libgfs2/misc.c +++ b/gfs2/libgfs2/misc.c @@ -248,42 +248,6 @@ void cleanup_metafs(struct gfs2_sbd *sdp) rmdir(sdp->metafs_path); } -static char *__get_sysfs(const char *fsname, const char *filename) -{ - char path[PATH_MAX]; - int fd, rv; - - memset(path, 0, PATH_MAX); - memset(sysfs_buf, 0, PAGE_SIZE); - snprintf(path, PATH_MAX - 1, "%s/%s/%s", SYS_BASE, fsname, filename); - - fd = open(path, O_RDONLY); - if (fd < 0) - return NULL; - rv = read(fd, sysfs_buf, PAGE_SIZE); - if (rv < 0) { - close(fd); - return NULL; - } - - close(fd); - return sysfs_buf; -} - -char *get_sysfs(const char *fsname, const char *filename) -{ - char *s; - char *p; - - s = __get_sysfs(fsname, filename); - if (!s) - return NULL; - p = strchr(s, '\n'); - if (p) - *p = '\0'; - return sysfs_buf; -} - int set_sysfs(const char *fsname, const char *filename, const char *val) { char path[PATH_MAX]; @@ -311,94 +275,6 @@ int set_sysfs(const char *fsname, const char *filename, const char *val) return 0; } -char *find_debugfs_mount(void) -{ - FILE *file; - char line[PATH_MAX]; - char device[PATH_MAX], type[PATH_MAX]; - char *path; - - file = fopen("/proc/mounts", "rt"); - if (!file) - return NULL; - - path = malloc(PATH_MAX); - if (!path) { - fclose(file); - return NULL; - } - while (fgets(line, PATH_MAX, file)) { - - if (sscanf(line, "%s %s %s", device, path, type) != 3) - continue; - if (!strcmp(type, "debugfs")) { - fclose(file); - return path; - } - } - - free(path); - fclose(file); - return NULL; -} - -/** - * mp2fsname - Find the name for a filesystem given its mountpoint - * - * We do this by iterating through gfs2 dirs in /sys/fs/gfs2/ looking for - * one where the "id" attribute matches the device id returned by stat for - * the mount point. The reason we go through all this is simple: the - * kernel's sysfs is named after the VFS s_id, not the device name. - * So it's perfectly legal to do something like this to simulate user - * conditions without the proper hardware: - * # rm /dev/sdb1 - * # mkdir /dev/cciss - * # mknod /dev/cciss/c0d0p1 b 8 17 - * # mount -tgfs2 /dev/cciss/c0d0p1 /mnt/gfs2 - * # gfs2_tool gettune /mnt/gfs2 - * In this example the tuning variables are in a directory named after the - * VFS s_id, which in this case would be /sys/fs/gfs2/sdb1/ - * - * Returns: the fsname - */ - -char *mp2fsname(char *mp) -{ - char device_id[PATH_MAX], *fsname = NULL; - struct stat statbuf; - DIR *d; - struct dirent *de; - char *id; - - if (stat(mp, &statbuf)) - return NULL; - - memset(device_id, 0, sizeof(device_id)); - sprintf(device_id, "%i:%i", major(statbuf.st_dev), - minor(statbuf.st_dev)); - - d = opendir(SYS_BASE); - if (!d) - return NULL; - - while ((de = readdir(d))) { - if (de->d_name[0] == '.') - continue; - - id = get_sysfs(de->d_name, "id"); - if (!id) - continue; - if (strcmp(id, device_id) == 0) { - fsname = strdup(de->d_name); - break; - } - } - - closedir(d); - - return fsname; -} - /* * get_random_bytes - Generate a series of random bytes using /dev/urandom. * diff --git a/gfs2/libgfs2/ondisk.c b/gfs2/libgfs2/ondisk.c index 73b2b0f..301d92d 100644 --- a/gfs2/libgfs2/ondisk.c +++ b/gfs2/libgfs2/ondisk.c @@ -415,21 +415,6 @@ void gfs2_dirent_out(struct gfs2_dirent *de, char *buf) memset(str->__pad, 0, sizeof(str->__pad)); } -void gfs2_dirent_print(struct gfs2_dirent *de, char *name) -{ - char buf[GFS2_FNAMESIZE + 1]; - - gfs2_inum_print(&de->de_inum); - pv(de, de_hash, "0x%.8X", NULL); - pv(de, de_rec_len, "%u", "0x%x"); - pv(de, de_name_len, "%u", "0x%x"); - pv(de, de_type, "%u", "0x%x"); - - memset(buf, 0, GFS2_FNAMESIZE + 1); - memcpy(buf, name, de->de_name_len); - print_it(" name", "%s", NULL, buf); -} - void gfs2_leaf_in(struct gfs2_leaf *lf, struct gfs2_buffer_head *bh) { struct gfs2_leaf *str = (struct gfs2_leaf *)bh->b_data; @@ -478,19 +463,6 @@ void gfs2_ea_header_in(struct gfs2_ea_header *ea, char *buf) ea->ea_num_ptrs = str->ea_num_ptrs; } -void gfs2_ea_header_out(struct gfs2_ea_header *ea, char *buf) -{ - struct gfs2_ea_header *str = (struct gfs2_ea_header *)buf; - - CPOUT_32(ea, str, ea_rec_len); - CPOUT_32(ea, str, ea_data_len); - str->ea_name_len = ea->ea_name_len; - str->ea_type = ea->ea_type; - str->ea_flags = ea->ea_flags; - str->ea_num_ptrs = ea->ea_num_ptrs; - str->__pad = 0; -} - void gfs2_ea_header_print(struct gfs2_ea_header *ea, char *name) { char buf[GFS2_EA_MAX_NAME_LEN + 1];

1 0

gfs2-utils: master - gfs2-utils: Only build group/ when gfs_controld is enabled
by Andrew Price 04 Jul '12

04 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=gfs2-utils.git;a=commitdiff;h=3b748462ec… Commit: 3b748462ec98cb5909b3ad9f3b639dbca38142fd Parent: 69b7a79e053aecaf7218134b8536bc0d7332a17e Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Wed Jul 4 12:30:15 2012 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Wed Jul 4 12:30:15 2012 +0100 gfs2-utils: Only build group/ when gfs_controld is enabled The previous commit only disabled gfs_controld from building when --enable-gfs_controld was omitted. When we're not building gfs_controld there really is no reason to build anything else in the group/ directory so this patch disables it by default. Signed-off-by: Andrew Price <anprice(a)redhat.com> --- Makefile.am | 6 +++++- group/Makefile.am | 6 +----- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/Makefile.am b/Makefile.am index 2e5d9ae..0af3e04 100644 --- a/Makefile.am +++ b/Makefile.am @@ -12,7 +12,11 @@ noinst_HEADERS = make/copyright.cf ACLOCAL_AMFLAGS = -I m4 -SUBDIRS = po group gfs2 doc +if BUILD_GFS_CONTROLD +DIR_GROUP = group +endif + +SUBDIRS = po $(DIR_GROUP) gfs2 doc install-exec-local: $(INSTALL) -d $(DESTDIR)/$(LOGDIR) diff --git a/group/Makefile.am b/group/Makefile.am index 2904ce1..5b7d0f8 100644 --- a/group/Makefile.am +++ b/group/Makefile.am @@ -1,7 +1,3 @@ MAINTAINERCLEANFILES = Makefile.in -if BUILD_GFS_CONTROLD -DIR_GFS_CONTROLD = gfs_controld -endif - -SUBDIRS = libgfscontrol gfs_control $(DIR_GFS_CONTROLD) man include +SUBDIRS = libgfscontrol gfs_control gfs_controld man include

1 0

gfs2-utils: master - gfs2-utils: Make building gfs_controld optional
by Andrew Price 04 Jul '12

04 Jul '12

Gitweb: http://git.fedorahosted.org/git/?p=gfs2-utils.git;a=commitdiff;h=69b7a79e05… Commit: 69b7a79e053aecaf7218134b8536bc0d7332a17e Parent: 46388807a62c24e4156584481f7df5eb7c8620cd Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Tue Jul 3 18:54:27 2012 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Tue Jul 3 19:02:53 2012 +0100 gfs2-utils: Make building gfs_controld optional As of Fedora 17 gfs_controld isn't required and also doesn't build. This patch adds an --enable-gfs_controld option to allow it to be built on older systems while defaulting to future norm of not building it. Signed-off-by: Andrew Price <anprice(a)redhat.com> --- README.build | 6 ++++++ configure.ac | 27 ++++++++++++++++++--------- group/Makefile.am | 6 +++++- 3 files changed, 29 insertions(+), 10 deletions(-) diff --git a/README.build b/README.build index 007f2d7..27e022a 100644 --- a/README.build +++ b/README.build @@ -21,6 +21,12 @@ Plus the following libraries: - libcfg - openaislib +By default gfs_controld is not built as it is not required in Fedora 17 and +later and cannot build on it. To re-enable it for older systems, use +--enable-gfs_controld when running the configure script (see below). The +library dependencies for gfs_controld are only required when this option is +given. + Run the following commands: ./autogen.sh diff --git a/configure.ac b/configure.ac index 570f826..d56cfac 100644 --- a/configure.ac +++ b/configure.ac @@ -95,6 +95,10 @@ AC_ARG_ENABLE([debug], [ --enable-debug enable debug build. ], [ default="no" ]) +AC_ARG_ENABLE([gfs_controld], + [ --enable-gfs_controld build gfs_controld. ], + [ default="no" ]) + AC_ARG_WITH([syslogfacility], [ --syslogfacility=FACILITY cluster default syslog facility. ], @@ -114,15 +118,20 @@ AC_ARG_WITH([kernel], KERNEL_CPPFLAGS="-I$KERNEL_DIR/include" -PKG_CHECK_MODULES([corosync],[corosync]) -PKG_CHECK_MODULES([cpg],[libcpg]) -PKG_CHECK_MODULES([sackpt],[libSaCkpt]) -PKG_CHECK_MODULES([logt],[liblogthread]) -PKG_CHECK_MODULES([ccs],[libccs]) -PKG_CHECK_MODULES([cfg],[libcfg]) -PKG_CHECK_MODULES([fenced],[libfenced]) -PKG_CHECK_MODULES([dlmcontrol],[libdlmcontrol]) -PKG_CHECK_MODULES([quorum],[libquorum]) +# gfs_controld isn't required in the latest versions of cluster +AM_CONDITIONAL([BUILD_GFS_CONTROLD], [test "x$enable_gfs_controld" = "xyes"]) +AS_IF([test "x$enable_gfs_controld" = "xyes"], [ + PKG_CHECK_MODULES([corosync],[corosync]) + PKG_CHECK_MODULES([cpg],[libcpg]) + PKG_CHECK_MODULES([sackpt],[libSaCkpt]) + PKG_CHECK_MODULES([logt],[liblogthread]) + PKG_CHECK_MODULES([ccs],[libccs]) + PKG_CHECK_MODULES([cfg],[libcfg]) + PKG_CHECK_MODULES([fenced],[libfenced]) + PKG_CHECK_MODULES([dlmcontrol],[libdlmcontrol]) + PKG_CHECK_MODULES([quorum],[libquorum]) +]) + PKG_CHECK_MODULES([zlib],[zlib]) # old versions of ncurses don't ship pkg-config files diff --git a/group/Makefile.am b/group/Makefile.am index 5b7d0f8..2904ce1 100644 --- a/group/Makefile.am +++ b/group/Makefile.am @@ -1,3 +1,7 @@ MAINTAINERCLEANFILES = Makefile.in -SUBDIRS = libgfscontrol gfs_control gfs_controld man include +if BUILD_GFS_CONTROLD +DIR_GFS_CONTROLD = gfs_controld +endif + +SUBDIRS = libgfscontrol gfs_control $(DIR_GFS_CONTROLD) man include

1 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

cluster-commits July 2012