Re: [squid-users] Issue with Digest and Number of Objects

Yes I agree with the major issue is the peers dropping offline. Is
there a way to keep this from happing without just turning off digest
all together. Being that it's saving a lot of icp queries.

Algorithm usage:
Cache Digest: 3602076 ( 97%)
Icp: 111327 ( 3%)
Total: 3713403 (100%)

Thanks
Dusten

On Tue, Dec 29, 2009 at 22:03, Amos Jeffries <squid3_at_treenet.co.nz> wrote:
> On Tue, 29 Dec 2009 14:11:44 -0500, Dusten Splan <dsplan_at_myyearbook.com>
> wrote:
>> Hi All,
>>   So I'm having an issue where squid will not cache more then 3843117
>> objects.
>
> NP: 2^24 objects limit per- cache_dir entry.
>
>>  Also on this same box we are seeing the traffic dip every
>> time it rebuilds the digest file.
>
> Some delay is expected while the 50MB digest is generated.
>
> I think probably the worse effect is that the peers are disappearing. This
> will most likely increase the network lag time for each request.
>
>>
>> Sample of log file.
>>
>> 2009/12/29 04:15:21.745| storeDigestRebuildStart: rebuild #276
>> 2009/12/29 04:15:21.745| storeDigestCalcCap: have: 38414447, want
>> 38414447 entries; limits: [8318
>> 0325, 103975384]
>> 2009/12/29 04:15:21.745| storeDigestResize: 81437985 -> 83180325;
>> change: 1742340 (2%)
>> 2009/12/29 04:15:21.745| storeDigestResize: small change, will not
> resize.
>> 2009/12/29 04:15:21.761| storeDigestRebuildStep: buckets: 8388608
>> entries to check: 16777220
>> 2009/12/29 04:15:36.125| storeDigestRebuildStep: buckets: 8388608
>> entries to check: 16777220
>> 2009/12/29 04:15:50| Detected DEAD Sibling: mypeer01
>> 2009/12/29 04:15:50| Detected DEAD Sibling: mypeer02
>> 2009/12/29 04:15:50| Detected DEAD Sibling: mypeer03
>> 2009/12/29 04:15:50.710| storeDigestRebuildStep: buckets: 8388608
>> entries to check: 16777220
>> 2009/12/29 04:15:54.956| storeDigestRebuildFinish: done.
>> 2009/12/29 04:15:55.010| storeDigestRewrite: start rewrite #276
>> 2009/12/29 04:15:55.010| storeDigestRewrite: url:
>> http://example.com/squid-interna
>> l-periodic/store_digest key: D412EC3C5799ADE5A3CF229E9BD27A16
>> 2009/12/29 04:15:55.040| storeDigestRewrite: entry expires on -1
>> (-1262078156)
>> 2009/12/29 04:15:55.065| storeDigestSwapOutStep: size: 50898741
>> offset: 0 chunk: 4096 bytes
>> 2009/12/29 04:15:55.079| storeDigestSwapOutStep: size: 50898741
>> offset: 4096 chunk: 4096 bytes
>> 2009/12/29 04:15:55.088| storeDigestSwapOutStep: size: 50898741
>> offset: 8192 chunk: 4096 bytes
>> ...
>> 2009/12/29 04:15:55.372| storeDigestSwapOutStep: size: 50898741
>> offset: 50888704 chunk: 4096 bytes
>> 2009/12/29 04:15:55.372| storeDigestSwapOutStep: size: 50898741
>> offset: 50892800 chunk: 4096 bytes
>> 2009/12/29 04:15:55.372| storeDigestSwapOutStep: size: 50898741
>> offset: 50896896 chunk: 1845 bytes
>> 2009/12/29 04:15:55.372| storeDigestRewriteFinish: digest expires at
>> -1 (-1262078156)
>> 2009/12/29 04:16:02| Detected REVIVED Sibling: mypeer01
>> 2009/12/29 04:16:02| Detected REVIVED Sibling: mypeer03
>> 2009/12/29 04:16:02| Detected REVIVED Sibling: mypeer02
>>
>>
>> Then start again an hour later.
>>
>> 2009/12/29 05:15:54.958| storeDigestRebuildStart: rebuild #277
>>
>>
>> Thanks
>>   Dusten
>
Received on Wed Dec 30 2009 - 20:22:07 MST