> To elaborate, squid should restart new url_rewrite_program instances
> when the number of live children falls to <= 50% of the configured
> number. So once 8 processes out of 15 die, squid should launch a whole
> new set of 15. You'll then have 23 url_rewriter processes, but squid
> will launch 15 more when they die and only 7 are left. The circle of
> life continues...
PS: yes that is a known bug. Squid should only start the 8 dead ones back.
It's on the queue for fixing one day.
>
> In squid 2.7, if you don't want to wait for half your helpers to die
> before squid launches more, you can adjust the threshold in
> helper.c:helperServerFree() by changing the following line:
>
> if (hlp->n_active <= hlp->n_to_start / 2) {
>
> And changing the right size of the evaluation from "/ 2" to, say, "* .
> 75" or similar.
>
> This can be used as a somewhat ghetto way of dealing with a rewriter
> that has a memory leak - put a dead-man's counter into the rewriter
> that causes the process to exit after X number of requests, and let
> squid launch new ones as needed. (Not that I've ever done anything
> like that myself, nosiree...)
>
Hmm, inspiration for a new idea and explanation of a bug in one post.
Thank you very much.
Amos
> -C
>
> On Mar 15, 2009, at 5:14 AM, Amos Jeffries wrote:
>
>> Dieter Bloms wrote:
>>> Hi,
>>> I use an url_rewrite_program, which seems to die after about 400000
>>> requests.
>>> Squid starts 15 processes, which are enough, but after some time one
>>> process after another die and at the end all processes where gone.
>>> Is it possible to let squid restart an url_rewrite_program, when it
>>> dies ?
>>
>> What version of Squid are you using that does not do this restart
>> automatically?
>> Squid only dies when ALL helpers for a needed service are dying too
>> fast to recover quickly.
>>
>> Amos
>> --
>> Please be using
>> Current Stable Squid 2.7.STABLE6 or 3.0.STABLE13
>> Current Beta Squid 3.1.0.6
>>
>
>
