Re: Can't Kill Squid Process

 unkillable process means problem blocks inside kernel.
 My bet would be buggy kernel threads implementation, especially
 that fsck after reboot hints that ext2 was corrupt already when
 squid hangs. Try diskd instead of async-io on linux.

 Other bet would be faulty hardware or non-ECC memory getting normal
 bit errors unnoticed.

On 28 Oct 2003 at 11:14, WA Support wrote:

> Hello,
>
> I have a Redhat Linux 7.1 system running kernel 2.4.18-26.7 under ext2
> type file system.
>
> I am running Version 3.0-PRE3-20030924 with '--enable-async-io'
> '--enable-basic-auth-helpers=NCSA' switches set at compile.
>
> I also have squidGuard-1.2.0 setup as a redirector.
>
> It runs fine for about a week normally, but then starts corrupting
> memory and gets to a point where I can't even kill the squid process.
> That is, '/etc/rc.d/init.d/squid stop' and 'kill -s 9' on the squid pid
> do not kill the process.
>
> I have to reboot my system to kill squid, but when I reboot, I always
> have to run e2fsck to fix corrputed inodes and data blocks.
>
> Does anyone know what might be going on here?
>
> Thanks,
> Murrah Boswell
>
Received on Wed Oct 29 2003 - 10:32:01 MST