Hello!
Relianoid 7.6
In the tes lab I found a strange bug in the backend maintenance mode.
In the test lab I Use 2 nodes Relianoid 7.6 in a cluster that balance Outlook client connections to Exchange.
When Outlook is connected to Exchange, I see the number of established connections in the farm to each Exchange backend (there are only 2 Exchange test servers in DAG).
I turn on maintenance mode on one of backends (cut), I see that client connections are reset, but after that they start reconnecting to the same backend, which is now in maintenance mode, that is, client connections should not be restored to it.
If you try to turn off and on maintenance mode on the backend, nothing will change, client connections will continue to be succesfully restored to the backend, which is in maintenance and should not accept connections intil the maintenance is disabled.
Maintenance mode only starts working if you turn it on on the backend and then restart the farm, after that client connections will not be restored to the backend, which is in maintenance mode.
Looks like a bug.
Farm is configured in l4xnat mode for port 443, number of backends 2.
Hi!
Could you please generate a support save (at System > Support Save) and share with us at: support AT relianoid DOT com ?
Thank you!
Hello!
I’ve send support file to you, check mail please )
Hi!
Thank you for the support save. It seems related to the persistence configured, as all the clients are Nated to the same IP address.
Please change the session persistence from “Source IP” to “Source Port”.
Looking forward to your feedback,
Cheers.
Hello!
I changed the session persistence from “Source IP” to “Source Port”, and maintenance mode is working better. In old versions ZEVENET maintenance mode working without changing session persistence from “Source IP” on “Source Port”. I want to understand maintenance logic is changed in RELIANOID balancer or I used not right settings always in ZEVENET?
Maintenance mode in ZEVENET always worked perfect, not depending of session persistence settings.
Problems started after I upgrade ZEVENET to RELIANOID.
Hi!
The maintenance mode and session persistence didn’t change at all.
In your infrastructure, were all the clients always NATed or it is a recent change? Cause if you had the clients load balanced to different backends using persistence by “Source IP” in the case where the clients are NATed then it was a Zevenet bug.
Just to confirm the correct behavior, please change it to “Source IP” and enable the debug at /usr/local/relianoid/config/global.conf , change from:
$debug=”0″;
to
$debug=”9″;
And then, reproduce the maintenance issue.
Finally, please share a fresh support save via email.
Kind Regards.
Hello!
Okay, thank you for information!
In my test ZEVENET-based infrastructure client connections always worked to NATed (lx4nat mode), interesting information about ZEVENET.
I dont need testing ZEVENET balancer because now I use RELIANOID.
RELIANOID good and simple balancer.
Thank you for help! =)
You’re welcome,
Enjoy RELIANOID!