Re: [squid-users] Workers reconfigure-question on 3.2.5

On Tue, Jan 8, 2013 at 10:28 AM, Tianyin Xu <tixu_at_cs.ucsd.edu> wrote:
> I tried your configuration but I don't have the problem. Could you
> paste your log file here?

In the log, I see, that kid5 is coordinator:
2013/01/08 11:15:46 kid5| Process Roles: coordinator

The following workers (workers 4) are now running (everything seems
working fine):
squid 28254 0.1 0.0 21052272 16032 ? S 11:15 0:00
(squid-coord-5) -f /etc/squid/squid.conf
squid 28255 0.1 0.0 21124980 16496 ? S 11:15 0:00
(squid-4) -f /etc/squid/squid.conf
squid 28256 0.1 0.0 21125112 19520 ? S 11:15 0:00
(squid-3) -f /etc/squid/squid.conf
squid 28257 0.1 0.0 21125404 20960 ? S 11:15 0:00
(squid-2) -f /etc/squid/squid.conf
squid 28258 0.1 0.0 21125096 16760 ? S 11:15 0:00
(squid-1) -f /etc/squid/squid.conf

After reconfigure, I got this error:
2013/01/08 11:15:11 kid5| /squid-cache05/swap.state: (2) No such file
or directory
FATAL: commonUfsDirOpenSwapLog: Failed to open swap log.
Squid Cache (Version 3.2.5): Terminated abnormally.

The error is OK, because there is no directory /squid-cache05.

>
> Is it possible that there are 4 kids but simply the naming has
> something wrong? e.g., kid2, kid3, kid4, kid5, etc?
With four workers configured, I have these processes:
squid 28254 0.1 0.0 21052272 16032 ? S 11:15 0:00
(squid-coord-5) -f /etc/squid/squid.conf
squid 28255 0.1 0.0 21124980 16496 ? S 11:15 0:00
(squid-4) -f /etc/squid/squid.conf
squid 28256 0.1 0.0 21125112 19520 ? S 11:15 0:00
(squid-3) -f /etc/squid/squid.conf
squid 28257 0.1 0.0 21125404 20960 ? S 11:15 0:00
(squid-2) -f /etc/squid/squid.conf
squid 28258 0.1 0.0 21125096 16760 ? S 11:15 0:00
(squid-1) -f /etc/squid/squid.conf

>
> Could you try to use
>
> cache_dir aufs /squid-cache00 100000 64 512
> cache_dir aufs /squid-cache01 100000 64 512
> cache_dir aufs /squid-cache02 100000 64 512
> cache_dir aufs /squid-cache03 100000 64 512

With the configuration above (your propose), every kid-process owns
ALL cache-dirs (shown with squidclient mgr:storedir). I think, this is
odd.

When I configure four cache_dirs with the variable ${process_number},
then one kid own one storedir -> this seems ok.

>
> instead of
>
> cache_dir aufs /squid-cache0${process_number} 100000 64 512
> cache_dir aufs /squid-cache0${process_number} 100000 64 512
> cache_dir aufs /squid-cache0${process_number} 100000 64 512
> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>
> ? (make sure your have /squid-cache00 -- /squid-cache03)

I tried beginning with squid-cache00, but I think there's not
squid-process 0, which access this volume. Isn't it beginning counting
with 1?

>
> I remember the counting of process_number do have some problem.
>
> -- Tianyin
>
> On Tue, Jan 8, 2013 at 1:17 AM, Tom Tom <tomtux007_at_gmail.com> wrote:
>> Hi
>>
>> I did run "squid -z" first with "workers 4". The necessary directories
>> were created successfully.
>>
>> I did not changed the workers in runtime, so I got no warning
>> concerning changing the workers.
>>
>> It makes no differences, if I use "squid -f /etc/squid/squid.conf -k
>> reconfigure" or "squidclient mgr:reconfigure" instead of "kill -1
>> $PID". One squid-kid (in my case kid5) always tries to access a
>> cache-directory after reconfigure. But the directory #5 does not
>> exist, because I just have 4 workers and for this number of workers, I
>> certainly have a cache_dir.
>>
>> How does others reconfigure their squid with workers running?
>>
>>
>>
>> On Tue, Jan 8, 2013 at 9:17 AM, Tianyin Xu <tixu_at_cs.ucsd.edu> wrote:
>>> Also, did you find the following log message in your logs:
>>>
>>> "WARNING: Changing 'workers' (from 1 to 4) is not supported and ignored"
>>>
>>> T
>>>
>>> On Tue, Jan 8, 2013 at 12:15 AM, Tianyin Xu <tixu_at_cs.ucsd.edu> wrote:
>>>> Hi, Tom,
>>>>
>>>> Do you have the directories "/squid-cache00" to "/squid-cache04" ready?
>>>>
>>>> Did you run $squid -z first?
>>>>
>>>> t
>>>>
>>>> On Mon, Jan 7, 2013 at 11:51 PM, Tom Tom <tomtux007_at_gmail.com> wrote:
>>>>> Hi
>>>>>
>>>>> I just added the following lines:
>>>>> workers 4
>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>>
>>>>> I tried the following reloads:
>>>>> - kill -1 on the squid main process -> no success.
>>>>> - kill -1 on the squid-coord-process -> no success.
>>>>>
>>>>> "No success" hereby means, that after the SIGHUP
>>>>> -> no 3128 listener was reachable
>>>>> -> a kid5 was tried to access a cache_dir /squid-cache05, which does not exist
>>>>>
>>>>> Kind regards,
>>>>> Tom
>>>>>
>>>>>
>>>>> On Tue, Jan 8, 2013 at 6:52 AM, Tianyin Xu <tixu_at_cs.ucsd.edu> wrote:
>>>>>> Hi, Tom,
>>>>>>
>>>>>> What did you change in the configuration file?
>>>>>>
>>>>>> t
>>>>>>
>>>>>>
>>>>>> On Mon, Jan 7, 2013 at 6:10 AM, Tom Tom <tomtux007_at_gmail.com> wrote:
>>>>>>>
>>>>>>> Hi
>>>>>>>
>>>>>>> I'm using 3.2.5 with four workers (workers 4). Each worker has its own
>>>>>>> cache_dir:
>>>>>>> workers 4
>>>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>>>> cache_dir aufs /squid-cache0${process_number} 100000 64 512
>>>>>>>
>>>>>>> When I now reconfigure squid with "squid -f /etc/squid/squid.conf -k
>>>>>>> reconfigure" or "squidclient mgr:reconfigure", then I got the
>>>>>>> following error in the cache.log:
>>>>>>> 2013/01/07 14:34:54 kid5| /squid-cache05/swap.state: (2) No such file
>>>>>>> or directory
>>>>>>>
>>>>>>> After this reconfigure, squidclient is no more usable (it always
>>>>>>> throws an 404). After this reconfigure, the main process for the
>>>>>>> 3128er-listener is then not the process "squid-coord-5" ->it's one of
>>>>>>> the kid-process.
>>>>>>>
>>>>>>> $ netstat -pantue | grep 3128
>>>>>>> tcp 0 0 0.0.0.0:3128 0.0.0.0:*
>>>>>>> LISTEN 0 1441481 20003/(squid-coord-5)
>>>>>>>
>>>>>>> $ squidclient mgr:reconfigure
>>>>>>>
>>>>>>> $ netstat -pantue | grep 3128
>>>>>>> tcp 0 0 0.0.0.0:3128 0.0.0.0:*
>>>>>>> LISTEN 0 1448779 20004/(squid-4)
>>>>>>>
>>>>>>> $ squidclient mgr:info
>>>>>>> HTTP/1.1 404 Not Found
>>>>>>> Server: squid
>>>>>>> Mime-Version: 1.0
>>>>>>> Date: Mon, 07 Jan 2013 13:58:37 GMT
>>>>>>> Content-Type: text/html
>>>>>>> Content-Length: 3299
>>>>>>> X-Squid-Error: ERR_INVALID_URL 0
>>>>>>> Vary: Accept-Language
>>>>>>> Content-Language: en
>>>>>>> ...
>>>>>>> ...
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Why does squid spawns a kid5-process (after the reconfigure) with a
>>>>>>> fifth cache_dir although there are just four configured? How can I
>>>>>>> reconfigure squid properly with workers running?
>>>>>>>
>>>>>>> Many thanks.
>>>>>>> Kind regards,
>>>>>>> Tom
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Tianyin XU,
>>>>>> http://cseweb.ucsd.edu/~tixu/
>>>>
>>>>
>>>>
>>>> --
>>>> Tianyin XU,
>>>> http://cseweb.ucsd.edu/~tixu/
>>>
>>>
>>>
>>> --
>>> Tianyin XU,
>>> http://cseweb.ucsd.edu/~tixu/
>
>
>
> --
> Tianyin XU,
> http://cseweb.ucsd.edu/~tixu/
Received on Tue Jan 08 2013 - 10:37:09 MST