Re: [squid-users] Failover & Load Balance

Ow Mun Heng wrote:

>In theory.. I think I need to experiment on this then.
>
>
You do not want to do this...I saw someone say you wanted to be on the
same subnet; I wouldn't even do it that way if it were me. I would run
serial cable direct or I would run crossover cable directly to
interfaces. The last thing you want is heartbeat flapping due to
latency. Usually heartbeat works well, but I have seen it bring up all
services/interfaces on all the servers which were failover for one
another. It is possible, and as you can imagine, it's not a case you
want a production environment to be in.

>Seeing that you're a sys-admin for managed hosting, do you have any tips
>on load-balance for multiple thousands of squid users?
>
>
We use LVS (Linux Virtual Server) as our loadbalancer at the moment.
www.ultramonkey.org and www.linuxvirtualserver.org

>What's the likelihood that squid will fail and then users are left w/o
>any I-net connections?
>
>
It's going to fail - it is just a matter of when.

Andrew

>
>
>>Andrew
>>
>>
>>>eg:
>>>
>>>---[center1]--interstate--[center2]---
>>>
>>>center1 and center2 will have independent squid installations. (but each
>>>will use the other as sibling)
>>>
>>>most likely using transparent proxy/wccp/ bridging (which one is better)
>>>and I would like users connected to center1's squid be fail-overed to
>>>center2's squid in case center1 goes down. (otherwise, being a
>>>tranparent proxy, if center1(or 2)'s squid goes down, (for some reason)
>>>I-net access for that center will fail.
>>>
>>>AFAIK, Hearbeat uses serial cables for monitoring, (I may be wrong, the
>>>last time I read the documentation).
>>>
>>>I would appreciate any help in this.
>>>
>>>--
>>>Ow Mun Heng
>>>Gentoo/Linux on D600 1.4Ghz
>>>Neuromancer 18:00:01 up 8:28, 6 users, 1.64, 0.76, 0.54
>>>
>>>
>>>
>
>--
>Ow Mun Heng
>Gentoo/Linux on D600 1.4Ghz
>Neuromancer 11:07:06 up 1:45, 6 users, 0.76, 0.54, 0.58
>
>
>
Received on Tue Nov 30 2004 - 06:11:26 MST