On 07/29/2013 03:49 PM, Martin Milata wrote:
Pushed, including the libreport patch.
I noticed that when processing a crash, following line appers in the
syslog (possibly several times):
abrt-server: Can't access '/var/tmp/abrt/post-create.lock': No such
file or directory
This is due to duplicate detection in abrt-handle-event which iterates
over all items in /var/tmp/abrt and fails to open the lock because it is
a broken symlink.
while ((dent = readdir(dir)) != NULL && crash_dump_dup_name == NULL)
char *dump_dir_name2 = concat_path_file(g_settings_dump_location,
dd = dd_opendir(dump_dir_name2, /*flags:*/ DD_FAIL_QUIETLY_ENOENT |
While the log message could be avoided by adding
DD_FAIL_QUIETLY_EACCESS to the dd_opendir (or explicitly skipping the
I am preparing a patch which will make DD_FAIL_QUIETLY_ENOENT
(which is already used there) to work as intended.
I was wondering why we don't store /proc/PID in the symlink
instead of just PID?
Then symlink will be a directory as far as stat() is concerned.
Probably an even worse situation :)