Re: [squid-users] Cache Replacement policies now working

From: Amos Jeffries <squid3_at_treenet.co.nz>
Date: Tue, 09 Sep 2008 00:15:10 +1200

Yasir Khan wrote:
> Thanks Amos,
> Currently I am using Version 3.0.STABLE2 of squid on RHEL5.

I'm not fully certain on it but I think the problem may be related to
another sen in the way squid files certain things in memory an disk. If
the diskd in your case is too slow saving an object before its
invalidated it can be left with a bad filename to deal with.

If AUFS is available on your system I'd try that on a later 3.0, before
STABLE5 things were pretty bad.

FWIW, I have some final adjustments then about to roll 3.0.STABLE9.

Amos

>
>
> regards
> Ali Yasir
>
> Amos Jeffries-2 wrote:
>> Yasir Khan wrote:
>>> Hi all,
>>> I have ten (10) squid-proxies and traffic is blanced using load balancer
>>> (Round Robin). The Problem seem to be of Cache Replacement Policy
>>> (Algorithem) as all the cache sizes are upto 90% filled. here are disk
>>> strucutre on eache cache server.
>>>
>>> Filesystem Size Used Avail Use% Mounted on
>>> /dev/cciss/c0d0p6 29G 22G 5.6G 80% /cache1
>>> /dev/cciss/c0d0p5 29G 22G 5.8G 79% /cache2
>>> /dev/cciss/c0d0p3 29G 22G 5.8G 79% /cache3
>>> /dev/cciss/c0d0p2 29G 21G 6.1G 78% /cache4
>>> ***********************************************
>>> and here is squid.conf configuration for allocating disk size to chache
>>>
>>> cache_swap_low 90
>>> cache_swap_high 95
>>>
>>> maximum_object_size 20480 KB
>>> maximum_object_size_in_memory 100 KB
>>>
>>> ipcache_size 10240
>>>
>>> cache_replacement_policy heap LFUDA
>>> memory_replacement_policy heap LFUDA
>>>
>>> cache_dir diskd /cache1 25000 16 256 Q1=72 Q2=64
>>> cache_dir diskd /cache2 25000 16 256 Q1=72 Q2=64
>>> cache_dir diskd /cache3 25000 16 256 Q1=72 Q2=64
>>> cache_dir diskd /cache4 25000 16 256 Q1=72 Q2=64
>>>
>>> ************************************************
>>> and here what i am getting in cache.log
>>>
>>> 2008/09/04 14:11:30| storeSwapOutFileClosed: dirno 3, swapfile 000807C6,
>>> errflag=FFFFFFFF
>>> 2008/09/04 14:11:30| (9) Bad file descriptor
>>> 2008/09/04 14:11:30| storeSwapOutFileClosed: dirno 3, swapfile 000807C7,
>>> errflag=FFFFFFFF
>>> 2008/09/04 14:11:30| (9) Bad file descriptor
>>> 2008/09/04 14:11:30| storeSwapOutFileClosed: dirno 3, swapfile 000807C1,
>>> errflag=FFFFFFFF
>>> 2008/09/04 14:11:30| (9) Bad file descriptor
>>> 2008/09/04 14:11:31| storeSwapOutFileClosed: dirno 3, swapfile 000807C2,
>>> errflag=FFFFFFFF
>>> 2008/09/04 14:11:31| (9) Bad file descriptor
>>> 2008/09/04 14:11:33| storeSwapOutFileClosed: dirno 3, swapfile 000807C3,
>>> errflag=FFFFFFFF
>>> 2008/09/04 14:11:33| (9) Bad file descriptor
>>>
>>>
>>> ********************************************************************
>>> Main problem is that there is no hits all the traffic is passing as it
>>> is..
>>> please help in resolving this issue ..
>>>
>>> regards
>>> Yasir Khan
>> Which squid version? On which operating system?
>>
>> Amos
>> --
>> Please use Squid 2.7.STABLE4 or 3.0.STABLE8
>>
>>
>

-- 
Please use Squid 2.7.STABLE4 or 3.0.STABLE8
Received on Mon Sep 08 2008 - 12:15:49 MDT

This archive was generated by hypermail 2.2.0 : Mon Sep 08 2008 - 12:00:02 MDT