Andrew Reid wrote:
> We've been having some problems with the current version of SQUID that
> we're using -- 2.3STABLE4. Basically, SQUID goes into a wierd mood every
> now and again, taking up 99.99% CPU and having a funny process status.
> In 'top', the process status shows up as 'R'. The thing can't be killed
> in any way at all. Believe me, we've tried everything. The only thing we
> can do to regain control of the box is to press the "Reset" button on
> the front of the case.
If you can't kill a process with "kill -9 <pid>" then there is a
kernel/driver problem somewhere. Possibly triggered by the process but it
is not the process that is at fault.
However, I have never seen processes get stuck in 'R' (Running). A process
in state 'R' is supposed to be killable. Are you sure it was 'R' and not
'D'?
> We've pretty much decided that we need to look at newer versions. 2.4 or
> even 2.3STABLE5 (if that fixes the aforementioned).
If it really is true that you cannot kill squid, not even with kill -9
<pid>, then upgrading probably won't help from this problem. But it may
save you from a number of other Squid problems.
> Essentially, we need the username and password quoted in the same style
> as is outputted to the access log. This may be a feature that should be
> able to be switched on and off through squid.conf.
It is a very good idea. Actually I have even made a patch to
Squid-2.2.STABLE3 once in a time but without an on/off option. (see
http://squid.sourceforge.net/hno/patches-devel.html#wait).
> So, is this a reasonable proposal? Is it something that is going to be
> too hard to implement?
It is reasonable.
Implementation is fairly trivial.
-- Henrik Nordström Squid HackerReceived on Fri Sep 07 2001 - 01:29:31 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:20 MST