On Monday 11 August 2003 18.51, Michael R. Wayne wrote:

> I suspect that much of the concern regarding RAID for the cache
> drives could be eliminated if Squid would properly "fail soft" once
> it saw N errors on a cache drive and simply ignore that drive from
> that point forward.

Squid partly does this already but you need to make sure the 
swap.state files are saved in a failsafe location as failures on 
these are fatal. There is a directive in squid.conf to tell where 
swap.state files should be saved.

In our cache products we also solve this by monitoring the drives and 
reconfiguring the system to not use failed drives in case of a 
failure. We also monitor the Squid process and redirect all requests 
to a secondary proxy in case the Squid process stops working 
properly.

Regards
Henrik Nordstr�m
MARA Systems AB, Sweden
http://www.marasystems.com/

Reply via email to