Well, for what its worth, I recompiled my kernel with 1024 file
descripters. I used the patch from
ftp://ftp.is.co.za/linux/local/kernel/filehandle.patch.linux.v8.01
I tried to set the things manually in the kernel, but I just couldn't get
it right. The shell and squid could not see 1024 file descripters. After
applying the patch and recompiling, it worked like a dream. My squid has
been up for a good part of a day and it is cooking. At the connection
rates I am getting now, it would have long sense hung and stoped. Here
are some stats for it for general interest.
Squid Object Cache: Version 1.1.22
Start Time: Tue, 04 Aug 1998 17:57:15 GMT
Current Time: Wed, 05 Aug 1998 12:08:50 GMT
Connection information for squid:
Number of TCP connections: 1073126
Number of UDP connections: 0
Connections per hour: 58985.5
Select loop called: 8150232 times, 8.036 ms avg
On Wed, 5 Aug 1998, Robin Bowes wrote:
> > Kelly wrote:
> > > I seem to get the same behavior to.. I am running with
> > > 1.1.20 and It is on redhat-5.1 on an alpha with 512 megs of
> > > ram, 39 gb of hd space.
> > > It seems to happen when my connections per second get about
> > > 23.5 to 25. I can send a HUP signal to the squid process
> > > and it is happy again.
> >
> > I'll see if that works for me. It doesn't really help since
> > it still
> > means that the cache is out of action until manually re-started.
>
> Nope. kill -HUP doesn't fix it.
Received on Wed Aug 05 1998 - 05:15:04 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:41:27 MST