plesk-daily statistics Segmentation fault

[GARMTECH]

After upgrade from 8.4 to 8.6 daily cron job exits with "Segmentation fault" error:

/etc/cron.daily/50plesk-daily:

/etc/cron.daily/50plesk-daily: line 4: 18584 Segmentation fault /usr/local/psa/admin/sbin/statistics >/dev/null 2>&1

How to fix that?

 

[ziba@]

Hello,

First of all make syre that all databases' directories in /var/lib/mysql/ are owned by mysql:mysql.

Another thing that might cause this is truncated or invalid symbol/sequence in a log file. Statistics fails on processing it and gives this error.

In any case you should execute statisitcs manually:
/usr/local/psa/admin/sbin/statistics

to check how it works and where it fails.
If database directories have correct owner and executing /usr/local/psa/admin/sbin/statistics does not give much of information then you can try using strace to debug:
strace /usr/local/psa/admin/sbin/statistics

and try to understand where it fails.

 

[JLChafardet]

Im having the same exact problem, its not related to mysql permissions, but it seems related to a bad character/line/record on certain domains logs, if you empty those logs, the statistics run smoothly.

this needs to be fixed asap, it is not directly related to awstats but at log information on domain.tld/statistics/logs

 

[GARMTECH]

strace-ing problem I found that this is due "MySQL server has gone away" error. Seems that it uses MySQL too much and our MySQL with strict limits drops connection after connection timeout or request processing time.

 

[embi]

I have the same problem, but don't think that it is related to MySQL.
Every second day I get the segfault. I then can manually execute the script again and it works. The next day it works too, but the day after it fails and the process begins again.

Any ideas?

 

[JLChafardet]

yes exactly like embi exposed it, i get segfault every 2nd or 3rd run of the script.

The problem may be related to the script as is, maybe something is wrong with it on plesk 8.6.0

 

[dash]

The reasons of statistics execution fails can be different from server to server.

So to diagnose the problem it is interesting to see output from statistics utility when it is running from console and fails.
#/usr/local/psa/admin/sbin/statistics

 

[Loup Artic]

Yep,

Same problem :

statistics[29363]: segfault at 00000000 eip 001d5563 esp bfae7b24 error 4
grsec: From 82.239.34.63: signal 11 sent to /usr/local/psa/admin/sbin/statistics[statistics:29363] uid/euid:0/0 gid/egid:0/0, parent /etc/cron.daily/50plesk-daily[50plesk-daily:29361] uid/euid:0/0 gid/egid:0/0

15 august at 5h20

And manuel :

statistics[31136]: segfault at 00000000 eip 001d5563 esp bfdb95c4 error 4
grsec: From 90.47.228.244: signal 11 sent to /usr/local/psa/admin/sbin/statistics[statistics:31136] uid/euid:0/0 gid/egid:0/0, parent /bin/bash[bash:31001] uid/euid:0/0 gid/egid:0/0

16 August at 13h.

Now, strace -l and im waiting.

 

[JLChafardet]

apparently, (at the very least on my personal particular case) the problem seemed to be related to a bug on openssh 5.0 p1 compilation, i just upgraded openssh to 5.1 p1 so i will keep you all posted.

 

[JLChafardet]

well it wasnt. im still getting the segfaults... so that wasnt, i read on wht that some segfaults were caused by a bug on sshd, and proceeded to upgrade sshd.... well that didnt fixed the segfaults.

 

[simpliq_adi]

Hello,

I have the same issue, I've tried everything, it looks like Plesk have a bug to fix.

 

[JLChafardet]

it seem so, the problem started with the latest plesk, with the older version I never got this segfaults.

 

[Loup Artic]

Now it's good after this manipulation.
But i have no idea for this problem.

Segfault only on a domain.
I execute statistics on the domain which trigger the segfault and it passed.

/usr/local/psa/admin/sbin/statistics --calculate-one --domain-name=blabla.com

Then, i re-do a global statistics and it passed too.

/usr/local/psa/admin/sbin/statistics --calculate-all

Now, no more segfault for the moment when doing global statistics.
To be continued...

(For info : 1-2K visits/day, 45-50K hits apache, small stats for the segfault ).

 

[Amin Taheri]

i only see this problem on Plesk 8.6 - and it is sporadic as it doesn't occur on the same machine twice in two days, so it appears some times but has occurred on every plesk 8.6 machine.

never occurred on 8.4

The error I see however is this:
Aug 19 05:31:47 server kernel: statistics[17480]: segfault at 0 ip 66129c45dc40 sp 714ffd5f3f48 error 4 in libc-2.5.so[66129c3e6000+14a000]

Aug 19 05:31:47 server kernel: grsec: signal 11 sent to /usr/local/psa/admin/sbin/statistics[statistics:17480] uid/euid:0/0 gid/egid:0/0, parent /etc/cron.daily/50plesk-daily[50plesk-daily:17477] uid/euid:0/0 gid/egid:0/0


Same error on aprox 20 machines, all centos5 x86_64

 

[embi]

Yes, you're right. It's sporadic and not compulsory every second day. I'm running Debian Etch.

 

[JLChafardet]

Yes is sporadic, im still working on finding a way to fix it myself, its not related to openssh nor mysql nor anything, it seems, parallels has some fixing to do.

 

[Amin Taheri]

One thing I did find is that its related to libc - not sure if that helps you at all.

 

[dedidude]

I get the same error in an e-mail from cron

/etc/cron.daily/50plesk-daily:

/etc/cron.daily/50plesk-daily: line 4: 24379 Segmentation fault /usr/local/psa/admin/sbin/statistics >/dev/null 2>&1

Server details:
Plesk Control Panel psa v8.6.0_build86080822.20 os_CentOS 4.2
Operating system Linux 2.6.9-67.0.22.ELsmp

 

[Amin Taheri]

This is now fixed in Plesk 8.6.0.2 - after installing I have confirmed that its no longer occuring when running stats

/usr/local/psa/admin/sbin/statistics --calculate-all

 

[CSN-Carl]

This is now fixed in Plesk 8.6.0.2 - after installing I have confirmed that its no longer occuring when running stats

/usr/local/psa/admin/sbin/statistics --calculate-all

8.6.0.2?

Not saw any announcement about this, do you have a change log?