Re: [squid-users] Load average problem

When squid receives a signal for reconfiguration
it restarts all ufdbGuard processes and it seems that the
newly started ufdbGuard processes rebuild the database.

I am the (biased) author of ufdbGuard.
ufdbGuard is faster, has more features and also
does not have the problem that is described
below because it has a different architecture.

-Marcus

Amos Jeffries wrote:
> Maxime Gaudreault wrote:
>> Hi list,
>>
>> I have a squid+squidguard server.
>>
>> When I make some changes on my "custom" blacklist I run these commands:
>>
>> /usr/bin/squidGuard -c /etc/squid3/squidGuard.conf.local -C all
>> /bin/chown -R proxy:proxy /var/lib/squidguard/db
>> /usr/sbin/squid3 -k reconfigure
>>
>> To apply the changes. My problem is that load average gets very high
>> during a long time. The faulty command is the last one: /usr/sbin/squid3
>> -k reconfigure
>>
>> Here is the timeline:
>>
>> I run "/usr/sbin/squid3 -k reconfigure"
>> After 50secs load average is 5
>> After 1m23 load average is 9.5
>> After 2m00 load average is 12
>> After 2m55 load average is 14
>> After 3m40 14.90 (peak)
>> It starts to get down 4m30sec later
>> After 4m55 load average is 12.2
>> After 5m35 load average is 10.45
>> After 5m45 load average is 8.85
>> After 6m00 load average is 6.89
>> After 6m30 load average is 4.17
>> After 7m00 load average is 2.53
>> After 7m30 load average is 1.53
>> After 8m00 load average is 0.93
>>
>> It takes 8 minutes before the load average gets back to 0. During this
>> time, CPU is not always 100%. Most of the time it's 0-10% with some
>> peak.
>>
>> I can't browse during these 8 minutes.
>>
>> Can someone help me ?
>
> What squid version?
> Under what normal connection load?
> with what configuration?
>
> Amos
Received on Sat May 23 2009 - 11:23:59 MDT