1.2b18 lockup?

Hi,

every now and then, my 1.2b18 locks up, doesn't service requests and cpu
usage increases substantially.
The machine load increases to about the amount of threads defined in
aiops.c., so I suspect it's some some of locking or race condition.
Shortly thereafter (10, 20 seconds), squid switches back to normal without
operator intervention. This is on Solaris 2.6.

I see 'WARNING - Async request queue growing: Length = 17' warnings in
cache.log. Increasing NUMTHREADS to whatever doesn't help, however; it just
makes the rest of the machine even less usable when squid decides to
lockup.

Any hints on this one ?

Thanks,

Markus

/////////////////////////////////////////////////////////////
  Markus Storm, mediaWays GmbH, Guetersloh, Germany
  Markus.Storm@mediaways.net
Received on Fri Mar 27 1998 - 01:16:41 MST