Re: [squid-users] Re: cache_peer problenms in accelerator mode

On 9/30/06, Henrik Nordstrom <henrik@henriknordstrom.net> wrote:
> fre 2006-09-29 klockan 22:19 -0700 skrev Edward Rosinzonsky:
>
> > I also see this in the cache.log file. Don't know what to make of it:
> >
> > 2006/09/29 21:35:25| temporary disabling (Not Found) digest from 127.0.0.1
>
> Either the parent hasn't built it's cache digest yet, or it doesnot have
> support for cache digests.

The "parent" is the original apache server.

>
> > 2006/09/29 21:36:32| temporary disabling (Service Unavailable) digest
> > from rele105
> > 2006/09/29 21:37:32| temporary disabling (Service Unavailable) digest
> > from rele107
>
> What are the above two peers "rele05" and "rele07"? Not mentioned
> before..

they are squid2 and squid3 respectively. I changed their names in the
email, and missed it here.

>
> > 2006/09/29 21:41:32| temporary disabling (Forbidden) digest from squid2
> > 2006/09/29 21:42:32| temporary disabling (Forbidden) digest from squid3
>
> Looks like the squid2 and squid3 does not allow digest exchanges.

What does that mean exactly? Both squid2 and squid3 are running squid 2.6.

>
> > 2006/09/29 21:47:27| Detected DEAD Sibling: squid2
> > 2006/09/29 21:47:27| Detected REVIVED Sibling: squid2
>
> Most likely too many ICP timeouts.
>
> If the servers are very close to each other then it's possible the
> automatic ICP timeout calculation is a bit too optimistic about the
> response time. Try setting "icp_query_timeout".
>

I tried icp_query_timeout 2000. That didn't help. Is there another
number I should use. The squids are on the same (gigabit) LAN, right
next to each other.

> > http_port 80 vhost
>
> Minor note: You should have a defaultsite=your.main.website as well, to
> support old clients not sending Host headers. Without this they will
> receive "Invalid request" as response as Squid can not determine which
> web site they requested.

thanks.
>
> Regards
> Henrik
>
>
>
Received on Sun Oct 01 2006 - 03:18:24 MDT