Re: [squid-users] Dead parent never reviving

Smells like a network issue.

Can you ping the remote office proxy from your parent?
(or the other way around if there is firewalls)

Do you run any kind of firewall between the two?
(including on the parent himself)

Things I may guess that could cause symptoms like this are

  *) A firewall triggering some kind of DoS protection rule.

  *) Local routing table hickup on the parent.

  *) Kernel bugs causing route caching bugs or arp errors on the
parent

  *) ARP hickup due to network errors. If the offices uses different
routers to connect to the LAN where your parent is located. Such
hickup can by example be caused by a rouge station trying to use the
same IP address as one of the remote office routers. Note: If the
remote offices all connect via the same router running as a router
(not proxy-arp/bridgeing), or if there is another LAN router
inbetween your parent and the connection point of your remote offices
then this CAN NOT be the cause.

Regards
Henrik

On Wednesday 04 September 2002 08.32, Eric Lawson wrote:

> I thought the detecting of a revived dead parent was supposed to be
> automatic? The parent cache in all these cases was still accessible
> by other child caches. I have an inkling that there was some
> network congestion on the link to the remote site. Could this be a
> factor? and if so, what can I do to resolve this so I don't have to
> reboot my parent cache all the time.
>
> All caches are running 2.3 Stable 4
>
> This is an extract from the cache log on the remote cache
>
> 2002/09/04 14:04:06| TCP connection to 10.65.9.132/3128 failed
> 2002/09/04 14:04:19| TCP connection to 10.65.9.132/3128 failed
Received on Wed Sep 04 2002 - 02:59:17 MDT