Segmentation fault (SEGV) in cyrus ptloader / CentOS 7

[Cornelius Hald]

Thanks for the info, so I think I'll stay with the "stable" version of
cyrus and just restart it once a night using cron. I found out that if I
prefork 10 ptloader processes I can get through a day ;)

Anyway, if you know how to get a stack trace I'd be glad to hear about
it.

BTW ctl_mboxlist seems to segfault on my system as well occasionally.

[...]
[493826.976159] ctl_mboxlist[25827]: segfault at 3 ip 00007f2d44c016e1
sp 00007fff1b6240d8 error 4 in libc-2.17.so[7f2d44a9f000+1b6000]
[...]

Cheers,
Conny


On Thu, 2014-11-20 at 16:54 +0100, Franz Skale wrote:
> Hi,
> i stepped back to version 2.5~dev2014082101-0~kolab1.
> Itried an upgrade to 2.5~dev2014111801-0~kolab1 which results in weird
> behaviour.
> Nevertheless, ctl_mboxlist didn't work with 2.5~dev2014082101-0~kolab1,
> so i took ctl_mboxlist from 2.5~dev2014111801-0~kolab1, which actually
> works !
> 
> 2.5~dev2014111801-0~kolab1 ctl_cyrusdb destroys the working mailboxes.db
> and finally fills up the imapd.log:
> cyrus-imapd/ctl_cyrusdb[6698]: IOERROR: skiplist RECSIZE not safe
> /var/lib/imap/mailboxes.db, offset 12192
> 
> Then i recreated the mailboxes.db using a backup, made before the
> upgrade, and was able to use ctl_mboxlist -d and cyr_dbtool skiplist show.
> When starting cyrus again, the mailboxes.db get corrupt again, showing
> the same error in imapd.log.
> 
> I think both versions are not bugfree, so i stay with my working, but
> mixed version.
> 
> Rgds.
> 
> Franz
> 
> 
> 
> 
> 
> 
> 
> 
> Am 19.11.14 um 15:15 schrieb Cornelius Hald:
> > On Wed, 2014-11-19 at 12:07 +1300, Torsten Grote wrote:
> >> On 11/19/2014 04:52 AM, Cornelius Hald wrote:
> >>> Unfortunately I do not get any of those gdb-backtrace file in /tmp
> >>>
> >>> Any ideas?
> >> Not really, but maybe this helps:
> >>
> >>     https://issues.kolab.org/show_bug.cgi?id=3652#c2
> > Thanks, but that's just about running a stand-alone program under gdb.
> > In my situation the cyrus master process has to start the ptloader child
> > processes under gdb. Which it does, but somehow wrong.
> >
> > Well, I'll keep trying...
> >
> > Cheers,
> > Conny
> >
> > _______________________________________________
> > users mailing list
> > users at lists.kolab.org
> > https://lists.kolab.org/mailman/listinfo/users
> 
> 
> _______________________________________________
> users mailing list
> users at lists.kolab.org
> https://lists.kolab.org/mailman/listinfo/users