processes dying

[jmackenz]

Hi

I'm running the following :
CPU

GenuineIntel, Intel(R) Xeon(R)CPU E5320 @ 1.86GHz

Version

Parallels Plesk Panel v10.4.4_build1013120126.11 os_CentOS 5

OS

Linux 2.6.18-028stab077.1
with 2gb Ram

Lately I've randomly been having processeses die/being killed

I've had named / qmail / plesk cp / spamd all randomly be dead

no reference in /var/log/messages as to why

any idea where to start looking? I'm guessing memory causing system to kill processes, but would think that a log reference would be added....

[prupert]

Looking at the kernel version (2.6.18-028stab077.1) you are on a Virtuozzo/OpenVZ container. You are probably running out of available memory. Of course, it is recommended to monitor memory usage on every server.

Determine if that is indeed the cause and investigate what is consuming the most memory and/or upgrade your memory.

[jmackenz]

on the host, or in the container? the 2gb is allocated to the container, the server itself has 8gb

[jmackenz]

and it's a vz container

[prupert]

on the host, or in the container? the 2gb is allocated to the container, the server itself has 4gb

Nobody here would be able to know of course. Monitor the memory usage on your server and you will know.

[jmackenz]

can you recommend a tool or a procedure for doing this monitoring.

I asked about the location as I assumed that if it was the container running low on memory that I'd be getting alerts from plesk, or errors written to the /var/log/messages , which i'm seeing neither

[jmackenz]

to follow up... this is what I've seen recently :

Code: Select all

May 28 15:06:43 cerberus kernel: kill_signal: 12 message(s) dropped
May 28 15:06:43 cerberus kernel: kill_signal(102.0): selecting to kill, queued 0, seq 2, exc 31954 0 goal 31954 0...
May 28 15:06:43 cerberus kernel: kill_signal(102.0): task ffff81021da0d5e0, thg ffff81021cbd9c40, sig 1
May 28 15:06:43 cerberus kernel: kill_signal(102.0): selected 1, signalled 1, queued 1, seq 2, exc 31954 0 red 84118 517
May 28 15:06:43 cerberus kernel: kill_signal(102.0): task ffff81021da0d5e0, thg ffff81021cbd9c40, sig 2
May 28 15:06:43 cerberus kernel: OOM killed process qmail-send (pid=25074, ve=102) exited, free=13994 gen=92.
May 28 15:06:43 cerberus kernel: OOM killed process qmail-smtpd (pid=59867, ve=102) exited, free=14146 gen=93.
May 28 15:06:43 cerberus kernel: OOM killed process qmail-smtpd (pid=57872, ve=102) exited, free=14899 gen=94.
May 28 15:06:43 cerberus kernel: OOM killed process couriertcpd (pid=25053, ve=102) exited, free=15085 gen=95.
May 28 15:06:43 cerberus kernel: OOM killed process imapd (pid=59861, ve=102) exited, free=16418 gen=96.
May 28 15:06:43 cerberus kernel: OOM killed process imapd (pid=57933, ve=102) exited, free=20200 gen=97.
May 28 15:06:43 cerberus kernel: OOM killed process udevd (pid=23189, ve=102) exited, free=21626 gen=98.
May 28 15:06:43 cerberus kernel: OOM killed process imapd (pid=57842, ve=102) exited, free=22062 gen=99.
May 28 15:06:43 cerberus kernel: OOM killed process qmail-smtpd (pid=60207, ve=102) exited, free=23950 gen=100.
May 28 15:06:43 cerberus kernel: OOM killed process python (pid=25303, ve=102) exited, free=24916 gen=101.
May 28 15:06:43 cerberus kernel: OOM killed process python (pid=25301, ve=102) exited, free=25412 gen=102.
May 28 15:06:43 cerberus kernel: OOM killed process pop3d (pid=60032, ve=102) exited, free=25802 gen=103.
May 28 15:06:43 cerberus kernel: OOM killed process qmail-smtpd (pid=60123, ve=102) exited, free=25895 gen=104.

does the free=x correlate to the amount of free memory? if so, is that bytes, kb , other unit?