Re: Squid fail using 1.0.20

From: Duane Wessels <wessels>
Date: Thu, 14 Nov 96 10:08:15 -0800 writes:

>Hi. Our Squid-1.0.20 running on Linux 2.0.23 died for the first time this
>afternoon. In fact it died twice :( I found this in cache.log
>FATAL: icpSendERROR: FD 23 unexpected port 3128.
>Squid Cache (Version 1.0.20): Terminated abnormally.

It could be indicative of some other problem. Like maybe your
Config data structure memory got trashed.

>I've also been getting buckets of these errors as well since I'm in error
>reporting mode :)
>[14/Nov/1996:19:37:30 +1300] comm.c:1235: commHandleRead: FD 21:
>read failure: (104) Connection reset by peer

These are more-or-less normal.

>I'm interested to know if that's anything nasty.
>And lastly... This one is interesting because we have a 1.2Gb cache spool
>which is only 25% full at the mo.
>[14/Nov/1996:15:56:22 +1300] store.c:1986: storeGetMemSpace:
>Allocating beyond the MAX Store with total size of 4199814
>[14/Nov/1996:15:56:22 +1300] store.c:1988: Start Deleting
>Behind for every pending objects
>[14/Nov/1996:15:56:22 +1300] store.c:1989: You should really
>adjust your cache_mem, high/low water mark,
>[14/Nov/1996:15:56:22 +1300] store.c:1990: max object size to
>suit your need.

This refers to the VM usage, not the disk usage. You have specified
4M for 'cache_mem' but squid is actively fetching objects which
exceed that amount.

Duane W.
Received on Thu Nov 14 1996 - 10:08:15 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:33 MST