Re: [squid-users] errors-to-log ratio/thresholds for software abort and comm_accept

From: Marc Elsen <marc.elsen@dont-contact.us>
Date: Fri, 25 Apr 2003 09:12:19 +0200

Adam wrote:
>
> Hello,
>
> Watching my logs these past two weeks since we rolled out our new 2.5STABLE2
> squid on an upgraded 2.8 Sun Ultra 60, I've seen "a lot" of warning/error
> type messages in the cache.log file. Researching these errors on this
> mailing list, the answer has often been "if you aren't getting too many"
> don't worry or "Sun's do that" (in the case of the s/w abort msg). So I am
> just wondering how many is "too many" and if my daily error message totals,
> relative to the daily size of the access.log file, are ok, average, or too
> high/a sign of a problem. The example below covers one full day and are
> representative of typical/average daily totals for the 2 errors and the
> access.log on a normal day. Any guidelines as to what is too many for each
> error are appreciated.
>
> thanks,
>
> Adam
>
> 886 messages with 'Software caused connection abort'
> 443 messages with 'comm_accept: FD *: (*) Software caused
> connection abort'
> 1,503,788 access.log = Total number lines in the /logs/access.log file

 These messages are usually caused by users (browsers) aborting
requests,
 (hitting stop button).
 So even in a 'that-many' condition observed, it is probably nothing
 to worry about.

 M.

-- 
 'Love is truth without any future.
 (M.E. 1997)
Received on Fri Apr 25 2003 - 01:12:57 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:15:29 MST