Re: V 1.2 or V 2.0 panics and dumps cache

At 19:59 11/10/98 +0200, you wrote:

>> Why does squid think cache1 is full, when the OS shows that it isn't?
>
>Bacause Squid failed to write to it.
>
>See you systems messages file, /var/log/messages for a possible
>explanation why writes failed to a disk that is not full. Possible
>causes range from hardware failures to OS bugs or even
>misconfigurations. To Squid all of them look the same.

Spot on Henrik.

Had a corrupted file system at about 83% on one of the 2
cache drives.

While e2fsck ing said drive, swap.state index file was found to
be corrupted, which meant the cache on said drive was dumped entirely.
(Is it possible to rebuild a missing swap.state with squid V2 ?)
So I took the opportunity to do some more thorough tests, which
proved the hard drive to be ok. (used badblocks in write mode)

(Suggestion for FAQ?)

In summary, if people see their cache dumping cache prior
to it being full, try the following : (Redhat 5.1)

1. Check inodes aren't all used up "df -i"
2. Examine /var/log/messages for OS or hardware related errors
3. Take each cache drive off-line, unmount, and test with e2fsck
4. If no problems are found with e2fsck, do badblocks testing.
   (Read-only for non-destructive or write mode for destructive.)

Thanks again for your help.

.----------------------------------------------.------------------------.
| Mark Reynolds <mark@rts.com.au> | Phone 08 9474 1211 |
| Network Manager, Reynolds Technology Pty Ltd | Fax 08 9474 4772 |
| 8 Preston Street Como 6152 Western Australia | Pager 08 9480 5884 |
| PO Box 120 Como 6952 Western Australia | http://www.rts.com.au/ |
`----------------------------------------------^------------------------'
Received on Tue Oct 20 1998 - 00:48:49 MDT