09-12-2013 08:52 AM
Hello all,
We are trying to put in production a new application but in our lab, a weird thing is happening.
We have an ACE20, FT configuration in one arm mode with several contexts and several serverfarms on each context.
In one of our contexts, we had configure 2 serverfarms, A and B, exactly the same. 4 realservers, 1 out of service and 3 inservice. The real servers are the same for both serverfarms. We had configured the same sticky and class map policies and the same access list. The only difference is the VIP that we use for the clients to reach the application.
The weird thing is that in serverfarm A, the ACE loadbalances perfectly. But in ACE B, 99% of the connections go to only one realserver in the serverfarm and the last 1% are distributed between the other 2 realservers in inservice state.
Any ideas why could this be happening?
Thanks in advance
Solved! Go to Solution.
09-15-2013 11:26 PM
Hi,
Few possible reasons are below :
The probes are failing for other servers from other load balancers.
Sticky connections pointing to one server.
Connection coming from a single ip ( IP based sticky)
regards,
Ajay Kumar
09-15-2013 11:26 PM
Hi,
Few possible reasons are below :
The probes are failing for other servers from other load balancers.
Sticky connections pointing to one server.
Connection coming from a single ip ( IP based sticky)
regards,
Ajay Kumar
09-16-2013 05:26 AM
Hi Ajay,
thanks for the response.
Regarding to your suggestions:
1) Probes, we verified and they are not failing. The realservers are permanently OPERATIONAL.
2) Sticky to the same server: do you configure that? My understanding was that sticky ip-netmask 255.255.255.255 will make every different IP client to go to a different server, not just only one. Not sure if there is something else that we should check to be sure that we do not have something misconfigured.
3) We are almost sure that they come from different IP addresses but that is something we are checking right now.
Thanks!
09-16-2013 05:48 AM
Hi,
If you are behind a firewall or a proxy then chances are that clients are coming with a single source IP address.
try using command :
show conn
regards,
Ajay Kumar
09-16-2013 07:13 AM
Hi Ajay,
we finally found that the app was receiving every request with the same IP. We are making changes on the app.
Thanks for response!
Fernando
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