> I'm using squid 1.1.20 on my linux box allocating 700Mb of cache_swap on
> one scsi disk with an Adaptec 1520 controller.
> After 30 days of regular work my Samsung scsi hd seems to have physical
> damages! So i've reinstalled my linux box with another scsci disk (now
> IBM and a more serious Adaptec 2930 controller). After 10 days the same
> problem!
> The error messages on syslog is:
>
> sense key Medium Error
> Additional sense indicates Unrecovered read error
> scsidisk I/O error: dev 08:02, sector ....
>
> What's the problem? Squid Cache too busy?
We had similar problems with Usenet on one of our machines;
we will killing Quantums at a rate of one every two months.
The solution? Better drives (IBM is a good choice, actually) and disk
striping/raid0 if you have that much activity. Linux should
support striping out of the box.
-- Gilles Melanson ViaNet Internet Solutions NOC System/Network Administrator 128 Larch St., Sudbury, ON, Can P3E 5J8 gilles@vianet.on.ca Phone#: (705) 675-0400 ICQ UIN 1944562Received on Mon Feb 09 1998 - 09:20:59 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:38:49 MST