01-11-2013 02:15 AM
Hi, I have a question regarding server-failover when using rserver+backup-rserver or serverfarm+backup-serverfarm (via sticky).
My understanding is, that in case a server fails, the existing connections will not fail over to a backup-server.
We are looking for a way to change that (default) behaviour - We've already searched the documentation etc. and found the "failaction reassign/purge". But when my colleague tried to have this confirmed via TAC, it got pretty confusing and we were told, there's no way to do/solve this..
So in a second attempt, does anyone have experience with this - maybe can confirm or give another solution?
Thanks & Regards,
Max
01-11-2013 02:59 AM
This was answered sometime back in few discussion:
first, not all servers accept TCP connection to be reassigned to them.
This is because tcp requires 3-way handshake and keeps track of sequence/ack numbers.Therefore, the servers need some way to exchange this information or to start in the middle.
On ace, we have 2 commands "failaction reassign" or "failaction purge".
The reassign is to simply move the connection to a backup server.
The purge is to kill the connection forcing the client to reopen a new one - this new connection will then be loadbalance to an active server.
01-11-2013 03:38 AM
Thanks, actually I read that discussion, but needed an confirmation that I'm on the right track, after Cisco TAC told us the exact opposite..
Thanks!
01-11-2013 04:13 AM
Thats the best possible answer for this. Good to know it was helpful.
with regards,
Ajay Kumar
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide