Cache corruption

Hi

I have a number of feature-stripped Redhat 6.0 based systems in
the field running squid 2.0.PATCH2.
Hardware is generally Pentuin 100 or better, 64Mb RAM, 6b IDE
HDD - about 50 workstations on the LAN.

Lately I have encountered a few problems where users shut the
Linux box down in an in-elegant fashion - that is, they just switch it
off instead of shutting down. To be fair, one or two of the boxes did
suffer power outages where the UPS did not last till the power was
restored (no UPS monitor software installed)

Anyway - the problem is that when the box reboots Squid throws a
wobbly due to file corruption. The filesystem heals itself, but squid
will not start up until the swap.state file is erased.
Question is: Does anyone have the same problem? Do you know
of a way that we can empower squid to heal itself or else rebuild
without losing too much cached data. I don't particularly want to
start the cache from scratch each time this happens, nor do I relish
the phone calls complaining that nothing is working.

BTW - the reason for 2.0.PATCH2 is that we are using Smartfilter
on these boxes and SF does not yet support newer squid versions.

Thank you all for maintaining this list and all the great solutions to
obscure problems.
Best Regards
Eugene Geldenhuys
MCNE ECNE MCSE MCP

TFX SOLUTIONS -
PROFESSIONAL NETWORK DESIGN ,IMPLEMENTATION AND SUPPORT
Received on Wed Sep 15 1999 - 04:32:43 MDT