3 messages in ru.sysoev.nginxRe: Hot Standby Hot Standby
FromSent OnAttachments
Stephen Nelson-SmithNov 21, 2008 2:03 am 
Almir KaricNov 21, 2008 2:31 am 
Stephen Nelson-SmithNov 21, 2008 4:43 am 
Actions with this message:
Paste this link in email or IM:
Paste this link in email or IM:
Atom feed for this thread
Paste this URL into your reader:
Subject:Re: Hot Standby Hot StandbyActions...
From:Almir Karic (alm@kiberpipa.org)
Date:Nov 21, 2008 2:31:20 am
List:ru.sysoev.nginx

stupid question, why your idea is any better than having 3 loadbalanced servers?

anyway, what you are asking can be done with recursive error_pages.

On Fri, Nov 21, 2008 at 10:03:56AM +0000, Stephen Nelson-Smith wrote:

Following on from testing yesterday, I can't see a way to do this:

* Traffic ---> backend1

* backend1 fails

* Traffic ---> backend2 (hot standby)

* backend2 fails

* Traffic ---> backend3 (2nd hot standby)

ie we cannot have a hot-standby-hot-standby.

The best I have come up with is weighting the two backup backends, so only a very small amount of traffic goes to one of the two live backup machines. This isn't ideal - the spec I need to meet is to have two tiers of failover, with the app only ever running on a single instance.

How can I achieve this? Is it possible?