Re: Squid connection retry problem -- fd leak?

From: Oskar Pearson <>
Date: Mon, 15 Sep 1997 16:52:56 +0200


We are having the same problem...

> > Hm...our server is getting ~20000 connection/hour. I used to get around
Ours is higher than this.

> > 30000 connection/hour during peak load. Output from netstat shows that our
> > server had around 200-300 tcp connections during that time. As I told you
> > below, most fds were hanging in writing state.
> <snip>

> Looks like a problem with not closing FD's. Try 1.NOVM.16. I had the same
> problem, although on Solaris 2.4, but it's over now with 1.NOVM.16.

We are running 1.NOVM.16 with the retry patch. It seems that we are running
50% higher on filehandles than with the VM version...

Unfortunately we can't run the VM version... perhaps once linux 2.0.31 is
out... currently linux 2.0.30 doesn't swap out fast enough, and I need the SYN
patches (just in case)

The NOVM version is substantially different...

Received on Mon Sep 15 1997 - 07:57:09 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:06 MST