Re: failover without downtime?

Amir Dafny-Man <[email protected]> writes:

> Regretfully this will not achieve what you want.

I thought not :-)

> At this point all existing connections to the original MASTER will be
> severed in an ugly way.

Yeah, that's what I feared.

> What you want is to place your service behind a load balancer that can
> gracefully stop existing traffic to one node while allowing new
> traffic to the second node.

We _do_ have a loadbalancer, but I want to fail over to the standby load
balancer :-)

Not sure if this ascii art survives (and it was never that great to
begin with) but this is essentially what we have:


vip ---- haproxy_1 ------ workerpool_1
                   \  /
                     \
                    / \
         haproxy_2 /----\ workerpool_2


keepalived master, haproxy_1, and workerpool_1 all run on machine A.
keepalived backup, haproxy_2, and workerpool_2 all run on machine B.

Given we now know we cannot fail over to haproxy_2 without downtime
we'll try another approach to debug the issue we are having.

Stig


------------------------------------------------------------------------------
Site24x7 APM Insight: Get Deep Visibility into Application Performance
APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month
Monitor end-to-end web transactions and take corrective actions now
Troubleshoot faster and improve end-user experience. Signup Now!
http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140