[squid-users] Squid 3.5.5 automatically reload itself in 2h rhythm

Amos Jeffries squid3 at treenet.co.nz
Thu Jun 18 09:23:18 UTC 2015


On 18/06/2015 8:54 p.m., Tom Tom wrote:
> Hi
> 
> Squid 3.5.5 on a SLES12 box with Rock and SSL-Bump enabled, reloads
> itself after every 2 hours (and only if there was initial some low
> traffic through it). Squid 3.3.13 on the same box doesn't reload
> itself after 2 hours.
> 
> In the cache.log are no suspicious entries. Everything looks and feels
> normal (the same log is shown at 06:31:xx and 08:31:xx -> every 2
> hours):
> 

If it was an assertion failure, or FATAL event, or expception error, or
transaction ERROR there should be at least one line above these ones you
quoted which explains the problem.

If there is no such lines then we can assume its a segmentation fault
and you need to seek down a core file (if any) which was produced. The
error details will be in there.


> 2015/06/18 10:31:24 kid1| Set Current Directory to /squid-cache01/
> 2015/06/18 10:31:24 kid1| Starting Squid Cache version 3.5.5 for
> x86_64-unknown-linux-gnu...
<snip>

> 
> Any hints for this behaviour?

If its regularly every 2 hours, I would look at your cron settings.
There may be some cronjob restarting or reconfiguring Squid every 2 hrs.

Another alternative is resource limits. If it takes more than 2hrs for
rock to load the cache index and the oom killer process of the kernel
kills the "resource intensive" Squid worker on or before completion you
would also see this type of behaviour.

Amos


More information about the squid-users mailing list