Forum Discussion
NimbostratusDiameter Traffic re-balancing
Hi all,
Following the last available deployment guide, I've configured the BIG-IP System for diameter traffic management.
I used iApp template using Universal persistence on Session ID and Round Robin balancing method; also, I added mblb profile to my VS.
All works fine but I've an issue when one (of two) pool members go down and return active.
In detail, when one server fail (ex. server-2) all its connection correctly go to the other node but when server-2 comes online, no new connections are re-balancing on it.
I also tried, without success, the "Action on service down" feature under pool configuration.
BIGIP version is 12.0 on VE.
Can someone help me?
Thanks in advance
5 Replies
After the pool member is restored, the new connection should be able to load balance to the restored server. Please file a case with F5 Support.
Jo_31162I have to explain myself better, new connections go on Server-2.
What I would like to know if it's possible to re-balance connections from Server-1 to Server-2.
For example, by iRule that detach from S-1 and then reselect S-2
Thank you
- Valentine_96813
I have not seen an irule that would do this outside or one using priority groups. You could change your LB method to something like least sessions. This would get you back faster as new connections would prefer the new server. Also, oneconnect could help out too. Short of this, the only way to shift the load back, would be to dump the connection/persit table.
- Jo_31162
Hi Valentine,
I tried least session also without success and OneConnect profile reuses idle connections if I'm not wrong.
Regardless of release, I thought something like this in order to disconnect from a server to make another load balancing decision:
https://support.f5.com/kb/en-us/solutions/public/9000/200/sol9278.html
- Jo_31162
I found following related thread also:
https://devcentral.f5.com/questions/fault-tolerant-long-lived-tcp-connections
Could someone tell me if there is a solution for this?
Rgds
