Re: [squid-users] WCCP Service Flapping every 40 minutse approx

From: Amos Jeffries <squid3_at_treenet.co.nz>
Date: Fri, 13 Nov 2009 18:15:04 +1300

Juan C. Crespo R. wrote:
> .Nov 12 15:14:51.724 VET: %WCCP-1-SERVICELOST: Service 80 lost on Web
> Cache x.x.x.x
> .Nov 12 15:15:06.764 VET: %WCCP-5-SERVICEFOUND: Service 80 acquired on
> Web Cache x.x.x.x
>
>
> Guys I'm having this very weird issue with just one of the servers.
>
> I have two servers set using, WCCP2 with 3 Routers, each router shows
> the same log, both server have the same Service ID, every 40 Minutes
> only one of them shows lost to the routers but SQUID nevers goes down?
>
> Could you send me any Idea?

Sure.

FYI: In the background Squid is sending its configured router a "HERE I
AM" message to indicate that its alive. The routers are watching for
this and update their alive/dead settings every send interval. If one is
missed they mark it down.

Which means...
  * Something on the network might be blocking the updates.
  * Something on the network might be delaying the updates.
  * router getting the update might not be passing it on for some reason.

Or Squid might be too busy to send the update on time and the routers
mark it down, then up again when Squid gets around to its send.

Squid can get 'to busy' when:
  * under very high network load
  * restarting
  * doing ache garbage collection on slow disks
  * garnage collecting a large chunk/count of objects
  * old squid with almost no IO happening could also lock up sending
stuff sometimes. (current releases since sometime mid-2.6 don't see this)

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE7 or 3.0.STABLE20
   Current Beta Squid 3.1.0.14
Received on Fri Nov 13 2009 - 05:15:42 MST

This archive was generated by hypermail 2.2.0 : Fri Nov 13 2009 - 12:00:04 MST