libteam October 2018

libteam@lists.fedorahosted.org

3 participants
5 discussions

[ISSUE] Failed to find "enabled" option.
by Xin Long 16 Sep '19

16 Sep '19

I found this err in "loadbalance" and "lacp" runner when adding ports. It's caused by trying to set "enabled" option in .port_link_changed() or .port_changed(). When a new port is added, the first 'port changed event' process is earlier than CMD TEAM_CMD_OPTIONS_GET, in this CMD, all the options are synchronized from kernel. It means there's no 'enabled' option yet when calling port_link_changed in the first 'port changed event' process. In lb_event_watch_port_link_changed and lacp_event_watch_port_changed, they call teamd_port_check_enable to set 'enabled' option. this err is triggered. I'm not sure why teamd_port_check_enable needs to check if 'enabled' option exists. I checked the ab's .port_link_changed(), it just sets it by calling team_set_port_enabled(), instead of checking 'enabled' option first. can we just use team_set_port_enabled to set it directly in .port(_link)_changed OR improve teamd_port_check_enable to avoid this err ? Thanks.

3 5

[PATCH] libteam/options: fix s32/u32 data storage on big endian
by Hangbin Liu 05 Nov '18

05 Nov '18

When put signed/unsigned int data to long on big endian(PPC64) and read it as singed/unsigned int, we will read from high bytes and get wrong number, e.g. wrong active port index and priority values. Fix it by using signed/unsiged int directly when store s32/u32 values. Signed-off-by: Hangbin Liu <liuhangbin(a)gmail.com> --- libteam/options.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/libteam/options.c b/libteam/options.c index 3f444cf..71cc99e 100644 --- a/libteam/options.c +++ b/libteam/options.c @@ -258,12 +258,13 @@ int get_options_handler(struct nl_msg *msg, void *arg) bool changed; int nla_type; int opt_type; - long tmp; - bool tmp_bool; void *data; int data_len = 0; int err; struct nlattr *data_attr; + unsigned int tmp_u32; + bool tmp_bool; + int tmp_s32; if (nla_parse_nested(option_attrs, TEAM_ATTR_OPTION_MAX, nl_option, NULL)) { @@ -304,8 +305,8 @@ int get_options_handler(struct nl_msg *msg, void *arg) switch (nla_type) { case NLA_U32: - tmp = (long) nla_get_u32(data_attr); - data = &tmp; + tmp_u32 = nla_get_u32(data_attr); + data = &tmp_u32; opt_type = TEAM_OPTION_TYPE_U32; break; case NLA_STRING: @@ -323,8 +324,8 @@ int get_options_handler(struct nl_msg *msg, void *arg) opt_type = TEAM_OPTION_TYPE_BOOL; break; case NLA_S32: - tmp = (long) nla_get_s32(data_attr); - data = &tmp; + tmp_s32 = nla_get_s32(data_attr); + data = &tmp_s32; opt_type = TEAM_OPTION_TYPE_S32; break; default: -- 2.5.5

2 1

[PATCH] teamd: add an option to force log output to stdout, stderr or syslog
by Hangbin Liu 05 Nov '18

05 Nov '18

By default, libdaemon prints the logs to stderr, and switches to syslog if the precess daemonizes. When some users, e.g. NetworkManager, run teamd foreground, the logs printed in syslog will as coming from NetworkManager, with mixed NetworkManager's own debug logs, which makes people feel a mess and hard to debug. Add option -l to support force teamd log output to stdout, stderr or syslog. Also add a global env TEAM_LOG_OUTPUT, which could be used for old version compatibility. Signed-off-by: Hangbin Liu <liuhangbin(a)gmail.com> --- man/teamd.8 | 3 +++ teamd/teamd.c | 18 +++++++++++++++++- teamd/teamd.h | 1 + 3 files changed, 21 insertions(+), 1 deletion(-) diff --git a/man/teamd.8 b/man/teamd.8 index 03b525c..3d27eae 100644 --- a/man/teamd.8 +++ b/man/teamd.8 @@ -69,6 +69,9 @@ Use the specified PID file. .B "\-g, \-\-debug" Turns on debugging messages. Repeating the option increases verbosity. .TP +.B "\-l, \-\-log-output" +Force teamd log output to stdout, stderr or syslog. +.TP .B "\-r, \-\-force-recreate" Force team device recreation in case it already exists. .TP diff --git a/teamd/teamd.c b/teamd/teamd.c index aac2511..6c47312 100644 --- a/teamd/teamd.c +++ b/teamd/teamd.c @@ -107,6 +107,7 @@ static void print_help(const struct teamd_context *ctx) { " file will be ignored)\n" " -p --pid-file=FILE Use the specified PID file\n" " -g --debug Increase verbosity\n" + " -l --log-output Force teamd log output to stdout, stderr or syslog\n" " -r --force-recreate Force team device recreation in case it\n" " already exists\n" " -o --take-over Take over the device if it already exists\n" @@ -140,6 +141,7 @@ static int parse_command_line(struct teamd_context *ctx, { "config", required_argument, NULL, 'c' }, { "pid-file", required_argument, NULL, 'p' }, { "debug", no_argument, NULL, 'g' }, + { "log-output", required_argument, NULL, 'l' }, { "force-recreate", no_argument, NULL, 'r' }, { "take-over", no_argument, NULL, 'o' }, { "no-quit-destroy", no_argument, NULL, 'N' }, @@ -152,7 +154,7 @@ static int parse_command_line(struct teamd_context *ctx, { NULL, 0, NULL, 0 } }; - while ((opt = getopt_long(argc, argv, "hdkevf:c:p:groNt:nDZ:Uu", + while ((opt = getopt_long(argc, argv, "hdkevf:c:p:gl:roNt:nDZ:Uu", long_options, NULL)) >= 0) { switch(opt) { @@ -191,6 +193,10 @@ static int parse_command_line(struct teamd_context *ctx, case 'g': ctx->debug++; break; + case 'l': + free(ctx->log_output); + ctx->log_output = strdup(optarg); + break; case 'r': ctx->force_recreate = true; break; @@ -1494,6 +1500,16 @@ static int teamd_start(struct teamd_context *ctx, enum teamd_exit_code *p_ret) /* Child */ } + ctx->log_output = ctx->log_output ? : getenv("TEAM_LOG_OUTPUT"); + if (ctx->log_output) { + if (strcmp(ctx->log_output, "stdout") == 0) + daemon_log_use = DAEMON_LOG_STDOUT; + else if (strcmp(ctx->log_output, "stderr") == 0) + daemon_log_use = DAEMON_LOG_STDERR; + else if (strcmp(ctx->log_output, "syslog") == 0) + daemon_log_use = DAEMON_LOG_SYSLOG; + } + if (daemon_close_all(-1) < 0) { teamd_log_err("Failed to close all file descriptors."); daemon_retval_send(errno); diff --git a/teamd/teamd.h b/teamd/teamd.h index 3934fc2..01bd022 100644 --- a/teamd/teamd.h +++ b/teamd/teamd.h @@ -99,6 +99,7 @@ struct teamd_context { enum teamd_command cmd; bool daemonize; unsigned int debug; + char * log_output; bool force_recreate; bool take_over; bool no_quit_destroy; -- 2.5.5

2 1

[patch libteam] teamd/teamd_runner_lacp: lacp_port_set_state
by MK Lim 05 Nov '18

05 Nov '18

if new state is PORT_STATE_DEFAULTED, set lacp_port->partner.state to 0x02 This ensures LACPDU is sent to update the far side when local local port transitions from default to current Currently, local port moves from default to current without sending LACPDU until 30sec later. This causes traffic disruption because far side has not completed negotiation Using 0x02 also cause LACPDU to be sent every 1 sec as long as port remains in defaulted state and LACP mode used is "active" --- teamd/teamd_runner_lacp.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/teamd/teamd_runner_lacp.c b/teamd/teamd_runner_lacp.c index 7b8f0a7..555aa06 100644 --- a/teamd/teamd_runner_lacp.c +++ b/teamd/teamd_runner_lacp.c @@ -965,7 +965,12 @@ static int lacp_port_set_state(struct lacp_port *lacp_port, case PORT_STATE_DEFAULTED: teamd_loop_callback_disable(lacp_port->ctx, LACP_TIMEOUT_CB_NAME, lacp_port); - /* fall through */ + memset(&lacp_port->partner, 0, sizeof(lacp_port->partner)); + lacp_port->partner.state |= INFO_STATE_LACP_TIMEOUT; + err = lacp_port_partner_update(lacp_port); + if (err) + return err; + break; case PORT_STATE_DISABLED: memset(&lacp_port->partner, 0, sizeof(lacp_port->partner)); err = lacp_port_partner_update(lacp_port); -- 2.15.1 (Apple Git-101)

3 4

Unable to get LACP runner working
by David Chen 25 Oct '18

25 Oct '18

Hi, I followed Redhat teaming configuration guide [1] to configure NIC teaming, except LACP runner, others work as expected. Here is my config file (from example lacp_1.conf): { "device": "team0", "runner": { "name": "lacp", "active": true, "fast_rate": true, "tx_hash": ["eth", "ipv4", "ipv6"] }, "link_watch": {"name": "ethtool"}, "ports": {"eth0": {}, "eth1": {}} } I can create LACP team interface, "teamdctl team0 state" shows correctly, but after I perform: $ ip link set up dev team0 $ ip a shows NO-CARRIER and state is DOWN for team interface. dmesg shows both eth0 and eth1 link are ready, but team0 is not. Also, I see messages saying, "eth0 port seems to be loop backed to the same team device". I am working on a POC project, please share if you have any suggestions or pointers for me to setup a working LACP runner. Thank you! [1] https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/ht…

2 2

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

libteam October 2018