Watchdog in CentOS7 Plesk 12.5 doesn't restart Apache httpd after a segfault

[Miguel_Tellería]

System: CentOS 7 with Plesk 12.5 #26

After a segfault of HTTPD for reasons similar to the situation described in KB 128431, the watchdog hasn't detected the server being down and has not done any restart reattempt or send an alert.

Any ideas of why this could have happened?

 

[Miguel_Tellería]

I forgot to add that the watchdog reacts correctly when we stop the httpd server manually.

 

[UFHH01]

Hi Miguel_Tellería,

consider to investigate your issue/problem with the help of "strace" ... for example:

First, please stop watchdog and apache completely and then restart apache with the command "strace -ff -o /var/log/test_trace.log -r apachectl start". Try several gracefull restarts ( and reloads ) and restart watchdog afterwards as well to investigate any issues/problems at your temporary log "/var/log/test_trace.log". If you need help with your investigations, please post the temporary log and include as well your apache2 modules, please.​

Additional manual for "strace" for CentOS 7 ( the commands don't vary for CentOS 6 or 5! ) : http://www.unix.com/man-page/centos/1/strace/ ( External link, please inform me if this link is dead, so I can provide another helpfull link! )

 

[Miguel_Tellería]

Hi UFHH01,

Thanks for your reply, it might be an interesting path to pursue.

However we need to find a moment when we won't disrupt our clients with tests. I'll keep your post handy if the issue comes up again.