Re: 1.1.14 continuous segmentation violation.

On Fri, 25 Jul 1997, Paul Rolfe wrote:
> I'm running 1.1.14 with no mods on a P166 with 128Mb RAM, Adaptec 2940UW,
> and a Micropolis 4.5Gb Wide SCSI drive, running Linux 2.0.27 with the 1024
> File Descriptors Patch.
>
> I'ts been running fine since the day of it's release, having only
> segfaulted once, and restarted itself straight away. However, last night,
> it died, and upon restart, continuously segfaulted until the number of
> tries in RunCache was reached. I reran RunCache, and the same problem
> persisted. Only a reboot of the machine fixed the problem.

Ah HA! I'm not alone! This describes the behaviour of my 1.1.14
precisely. It ran fine for a couple of days, and then I got sprays of
"Poor Squid is dead" messages in my e-mail.

> 97/07/24 20:06:06| Configuring Sibling proxy2.omen.com.au/8080/3130
> FATAL: Received Segment Violation...dying.
> 97/07/24 20:06:06| storeWriteCleanLog: Not currently OK to rewrite swap log.
> 97/07/24 20:06:06| storeWriteCleanLog: Operation aborted.

I don't think it's related to the sibling configuration:

97/07/23 02:18:23| ERR_CONNECT_FAIL: http://205.228.184.30/FIDO-1/BID-43/NumItm-10/14024961-15480983
FATAL: Received Segment Violation...dying.
97/07/23 02:50:39| storeWriteCleanLog: Starting...

As you can see, it was up and running for about 30 minutes before it
crashed here. The second time it crashed was immediately after startup,
while it was still reading the log file. The third time was immediately
after startup after configuring a sibling (I was testing with teh
squid.nlanr.net server) but before having a chance to read the log file,
but that's probably because it didn't write one the previous time it
crashed.

It died two or three more times after that, all immediately after startup,
and finally it stayed up until the following morning, but that's prbably
because I stopped using it. It crashed again shortly after 9:00am when
the users started using it. It's been up ever since I switched back to
1.1.10+retry.

At least this problem is less "in-your-face" than 1.1.12 or 1.1.13's were.

        -Mike Pelletier.
Received on Fri Jul 25 1997 - 06:39:53 MDT