Hello everyone,
We've upgraded 2 environments (DEV and TEST) to 16.1.0 and we've noticed that the admin tower command no longer works as expected, and the servers and services display strangely in the NSA.
I'll start with the NSA. When we login we sometimes see all servers in the cluster, sometimes none, and sometimes just a couple of them. The same is true of the services. On the individual servers the services show as running (
service status all gives the expected results), and the load-balancer is showing all app services as healthy. Clarity seems to be working OK though.
If we execute the
admin tower command set set
trace on, we don't see anything on any of the app servers. Similarly the
list clients option just shows "none", even after explicitly setting
autodiscovery on.
These used to work fine before we upgraded, although I'll admit that the NSA behaviour of now you see it / now you don't is a feature I've seen many times before. It does seem to have worsened though.
Another thing we've seen is that the cmn_jgroups_ping table isn't being populated (we can't use multicast so JDBCPing is in use).
I've looked in the release note but didn't see anything that might explain the above. In the beacon-system.log on all servers the "Validated by server" messages show for all 4 servers in the cluster so I think things are sort of working.
Has anyone else seen this behaviour?
In closing, everything is fine in production (which is still at 16.0.3).
As always any and all suggestions will be gratefully received.
Paul