[dev.icinga.com #6588] Notifications causing segfault from exim

[icinga-migration]

This issue has been migrated from Redmine: https://dev.icinga.com/issues/6588

Created by dseeley on 2014-06-26 13:43:13 +00:00

Assignee: gbeutner
Status: Resolved (closed on 2014-06-27 06:20:03 +00:00)
Target Version: 2.0.1
Last Update: 2014-06-27 06:20:03 +00:00 (in Redmine)

Icinga Version: v2.0.0-19-g5518826

---

Ubuntu 14.04 3.13.0-30-generic, x86_64
icinga2, icinga2-classicui (v2.0.0-19-g5518826 from snapshots, or 2.0.0 from releases)
exim4-deamon-light 4.82-3ubuntu2 (also tried -heavy, same problem)

mail-host-notification.sh can be successfully run from the command line as a normal user (removing the environment vars), but there is a segfault from exim when it is called from icinga2.

With ssmtp, there is no such issue (not tried postfix).

According to /var/log/icinga2/debug.log, notifications are sent.

[2014-06-26 06:35:01 +0100] information/NotificationComponent: Sending reminder notification for object 'TEST2'
[2014-06-26 06:35:01 +0100] debug/Notification: FType=32, TypeFilter=511
[2014-06-26 06:35:01 +0100] information/Notification: Sending notification for user 'icingaadmin'
[2014-06-26 06:35:01 +0100] notice/CheckerComponent: Pending checkables: 0; Idle checkables: 25; Checks/s: 0.8
[2014-06-26 06:35:01 +0100] notice/Process: Running command '/etc/icinga2/scripts/mail-host-notification.sh': PID 15881
[2014-06-26 06:35:01 +0100] information/Notification: Completed sending notification for object 'TEST2'
[2014-06-26 06:35:02 +0100] notice/Process: PID 15881 terminated with exit code 0

However, notifications are not processed by exim, and in /var/log/syslog, there is a segfault from exim at the time the notification is sent, e.g.:

Jun 26 06:35:01 xxx-icinga kernel: [508955.594811] sendmail[15886]: segfault at 7fff53ab12d8 ip 00007fd8c1b7825a sp 00007fff53ab12c0 error 6 in exim4[7fd8c1b65000+e6000]

This is possibly a bug in exim, but there might be a subtlety in the way that it is called that can be worked around within icinga2?

Changesets

2014-06-27 06:18:32 +00:00 by gbeutner 62896cb

Increase default stack size

fixes #6588

[icinga-migration]

Updated by gbeutner on 2014-06-26 13:51:28 +00:00

Hm, that's interesting. Try running "ulimit -s 64" on your shell and see if you can reproduce the crash by running the script.

[icinga-migration]

Updated by gbeutner on 2014-06-26 13:51:36 +00:00

• Status changed from New to Feedback
• Target Version set to 2.0.1

[icinga-migration]

Updated by dseeley on 2014-06-26 13:58:26 +00:00

Yes, indeed, this does reproduce the issue.

[icinga-migration]

Updated by gbeutner on 2014-06-26 14:05:18 +00:00

• Assigned to set to gbeutner

[icinga-migration]

Updated by gbeutner on 2014-06-26 14:07:01 +00:00

Ok, can you test this with 256, 512, 1024, 2048, etc. and figure out the smallest value that doesn't break exim?

[icinga-migration]

Updated by dseeley on 2014-06-26 15:11:59 +00:00

At 523, it fails, at 524, it succeeds.

[icinga-migration]

Updated by gbeutner on 2014-06-27 06:20:04 +00:00

• Status changed from Feedback to Resolved
• Done % changed from 0 to 100

Applied in changeset 62896cb.