[squid-users] stoping after rotate

[Marcus Kool]

On 09/08/2015 08:11 AM, Jorgeley Junior wrote:
> Thank you all, this is the output:
> vm.overcommit_memory = 0
> vm.swappiness = 60
> I have a Redhat 6.6

The value of vm.overcommit_memory is OK.
The default value for vm.swappiness is way too high. It means that Linux swaps out parts of processes when they are idle for a while.
For better overall system performance, you want those processes in memory as long as possible and not swapped out so I recommend to change it to 15.
This implies that the OS has 15% of the physical memory available for file system buffers which is plenty.

You only mentioned that the swap is 32 GB.  What is the size of the physical memory ?

Did you already increase the swap ?

Marcus


> 2015-09-05 15:08 GMT-03:00 Marcus Kool <marcus.kool at urlfilterdb.com <mailto:marcus.kool at urlfilterdb.com>>:
>
>     On Linux, an important sysctl parameter that determines how Linux behaves with respect to VM allocation is vm.overcommit_memory (should be 0).
>     And vm.swappiness is important to tune servers (should be 10-15).
>
>     Which version of Linux do you have and what is the output of
>         sysctl -a | grep -e vm.overcommit_memory -e  vm.swappiness
>
>     Marcus
>
>
>     On 09/04/2015 07:04 PM, Jorgeley Junior wrote:
>
>         Thanks Amos, i will increase the swap
>
>         Em 04/09/2015 17:22, "Amos Jeffries" <squid3 at treenet.co.nz <mailto:squid3 at treenet.co.nz> <mailto:squid3 at treenet.co.nz <mailto:squid3 at treenet.co.nz>>> escreveu:
>
>              On 5/09/2015 7:16 a.m., Jorgeley Junior wrote:
>               > Thanks Amos, my swap is 32GB, so that's causing the error as you said.
>               > Which is the better choice: increase the swap size or reduce the
>               > cache_mem???
>               >
>
>              Both probably. 128 GB swap I suspect you will need.
>
>              Increase the swap so the system lets Squid use more virtual memory.
>
>              Decrease the cache_mem so that Squid does not actually end up using the
>              swap for its main worker processes. That is a real killer for performance.
>
>
>              Amos
>
>
>
>         _______________________________________________
>         squid-users mailing list
>         squid-users at lists.squid-cache.org <mailto:squid-users at lists.squid-cache.org>
>         http://lists.squid-cache.org/listinfo/squid-users
>
>
>
>
> --
> *_
> _*
> *_
> _*