[squid-users] High utilization of CPU squid-3.5.23, squid-3.5.24

[Vitaly Lavrov]

On 02.02.2017 00:45, Yuri Voinov wrote:
> Yes, it is require to perform extended diagnostics. Including the system
> level.
>
> BTW, it can also network IO. And, it is possible that even a slow DNS.
> Have to search.

squid-3.5.23 or squid-3.5.24 with bug-4606-v3.patch (http://bugs.squid-cache.org/show_bug.cgi?id=4606)

All cache_dir are located on SSD. Read-write up to 2 mbytes/s. IO wait < 3% - no problem.
Network average 20-40Mbit/s peak 110Mbit/s ( gigabit network ). 60-100 request/s
2 local dns-server.

This is very similar to http://bugs.squid-cache.org/show_bug.cgi?id=4477 (no solution)

>
> 02.02.2017 3:34, Eliezer Croitoru пишет:
>> I believe that the squid manager info page should give some clue about the number of concurrent requests.
>>
>> If it's above some number(300-400 and above) per second then removing the cache_dir from the server for a windows of a day will answer if it's a DISK IO bottle neck or something else.
>>
>> All The Bests,
>> Eliezer
>>
>> ----
>> Eliezer Croitoru
>> Linux System Administrator
>> Mobile: +972-5-28704261
>> Email: eliezer at ngtech.co.il
>>
>>
>> -----Original Message-----
>> From: squid-users [mailto:squid-users-bounces at lists.squid-cache.org] On Behalf Of Vitaly Lavrov
>> Sent: Wednesday, February 1, 2017 10:56 PM
>> To: squid-users at squid-cache.org
>> Subject: [squid-users] High utilization of CPU squid-3.5.23,squid-3.5.24
>>
>> Periodically squid begins to linearly increase the use of the CPU.
>> Sometimes this process reaches 100%. At random moment of time the CPU usage is reduced to 5-15%, and in the presence of client requests can again start linearly increasing use of CPU.
>>
>> In the protocols are no error messages.
>>
>> CPU consumption does not correlate with the number of requests and traffic.
>>
>> The increase CPU consumption from 0 to 60% occurs in about 4-5 hours, and to 100% for 6-8 hours.
>>
>> A typical graph of CPU usage can be viewed on http://devel.aanet.ru/tmp/squid-cpu-x.png
>>
>> With the "perf record -p` pgrep -f squid-1` - sleep 30" I have received the following information:
>>
>> At 100% CPU load most of the time took 3 calls
>>
>>   49.15% squid squid [.] MemObject :: dump
>>   25.11% squid squid [.] Mem_hdr :: freeDataUpto
>>   20.03% squid squid [.] Mem_hdr :: copy
>>
>> When loading CPU 30-60% most of the time took 3 calls
>>
>>   37.26% squid squid [.] Mem_node :: dataRange
>>   22.61% squid squid [.] Mem_hdr :: NodeCompare
>>   17.31% squid squid [.] Mem_hdr :: freeDataUpto
>>
>> What is it ? Is it possible to somehow fix it?
>>
>> System: slackware64 14.2
>>
>> sslbump not used. http only.
>>
>> Part of config:
>>
>> memory_pools off
>> memory_pools_limit 512 MB
>> cache_mem 768 MB
>> maximum_object_size_in_memory 64 KB
>> cache_dir ufs           /cache/sq_c1 16312 16 256
>> cache_dir ufs           /cache/sq_c2 16312 16 256
>> cache_dir ufs           /cache/sq_c3 16312 16 256
>> _______________________________________________
>> squid-users mailing list
>> squid-users at lists.squid-cache.org
>> http://lists.squid-cache.org/listinfo/squid-users
>>
>> _______________________________________________
>> squid-users mailing list
>> squid-users at lists.squid-cache.org
>> http://lists.squid-cache.org/listinfo/squid-users
>
>
>
> _______________________________________________
> squid-users mailing list
> squid-users at lists.squid-cache.org
> http://lists.squid-cache.org/listinfo/squid-users
>