Re: pinger.c bug?

Hi,

i've written this to this list last week, but nobody responded then:

When I send a kill -HUP (kill -1), or a "squid -k reconfigure", to the
running squid-2.1-PATCH1, squid restarts quickly
and then often writes the line "Cache Dir #0/#1/#2 (DIRTY)" in cache.log
Then a restart takes rather long as it checks the cache-dir's.

It seems like squid forgets to "cleanly shutdown" or at least forgets
to mark the cache-dir as properly closed. (There's a sort of semaphore
file for this purpose ain't it: swap.state.last-clean)

In squid-2.0RELEASE I've never noticed this behaveour, so in my opinion
the shutdown/restart routine in squid-2.1.PATCH1 is broken.

Regards, Eddy
Network Specialist
http://www.hzeeland.nl/~eddy/

>(I've upgraded to a patch1 binary now, that fixed the pinger.c bug).
>thanks.
>
>I usually do a kill -TERM ${PID} with an init script i have. is that
>wrong?. There is another guy who also does some squid stuff too, I think
>he may signal 9 it sometimes.
>
>What is the correct way to shut it down?.. as there are usually two squid
>processes to kill..
>
>TIA,
>
>J.
>
>On Fri, 27 Nov 1998, Henrik Nordstrom wrote:
>
>> > the line "Cache Dir #0/#1/#2 (DIRTY)" shows up each time i
>> > start squid.. is that a bad thing?.
>>
>> Could be a bad thing. Squid recovers automatically, but the performace
>> is sluggish during the recovery, and things may break more than usual
>> (activity during rebuild is not very much tested).
>>
>> How do you stop your Squid?
>>
>> ---
>> Henrik Nordstrom
>> Spare time Squid hacker
>>
Received on Mon Nov 30 1998 - 13:34:58 MST