Problems with Squid 1.1.20

We're experiencing strange behaviour on one of our Squids - the rest are
working fine. We're running 4 Squids, all on Alpha, all the same version on
the same version of Digital Unix. On one machine it doesn't work properly.
It doesn't seem to matter what version of the OS or Squid either, it seems
to be just the machine. Even stranger is that it used to work fine, then one
day it just stopped.
The problem we're seeing is that all of a sudden Squid stops responding to
connections. Squid continues to run, but no longer responds to connections -
they time out. In addition once Squid has gone into this state it won't shut
down properly when killed. It seems to stay in the waiting for active
connections to finish state.

Below are the contents of cache.log last time it happened.
The kill was done at Mon May 4 15:39:20 NZST 1998, so the times on the last
two lines of the log are interesting - why didn't the time change?

1998/05/04 15:32:47| storeWriteCleanLog: Not currently OK to rewrite swap log.
1998/05/04 15:32:47| storeWriteCleanLog: Operation aborted.
1998/05/04 15:32:47| storeRotateLog: Rotating.
1998/05/04 15:32:47| stat_rotate_log: Rotating
1998/05/04 15:32:47| Preparing for shutdown after 5679 connections
1998/05/04 15:32:47| Waiting 30 seconds for active connections to finish

We are rotating the logs every 10 minutes or so for accounting purposes.

Is this a known problem with Squid? If not, what can I do to track down what's
happening quickly?

Thanks,

Brent Foster
Massey University, Palmerston North, New Zealand
Received on Tue May 05 1998 - 20:51:53 MDT