NSX Controller Nodes going not connected 6.3

[vmmedmed]

A co-worker at a microsegmentation gig pointed out that Controller Nodes frequently go to disconnected.

In Networking and Security I'm referring to Installation and Upgrade/Management/Controller Nodes.

I only have limited access and no ability to open a ticket on the environment. I wonder if anyone

on the forum had seen this before. They have many controller clusters - 10 or more whereas in

my previous environment we only had one and I've never seen this issue. The problem it

causes is when we want to publish distributed firewall rules - the rules will fail to push to the

clusters that are temporarily in disconnected state. The condition is always very transient

and comes back in a few moments. But it can cause a problem with publishing of the rules.

Has anyone ever seen this before? Is it a known bug? Other thought?

[mauricioamorim · Accepted solution]

Disconnected shows on controllers when they lose IPSEC tunnel with other controllers. You can check this with command “show control-cluster network ipsec tunnels“. One common reason for this is disk I/O latency. Check if there are no such alerts with command “show disk-latency-alert” or in the GUI in Actions > Fetch Disk Alerts on the Installation and Upgrade screen while selecting a controller.

---------------------------------------------------------------------------------------------------------

Was it helpful? Let us know by completing this short survey here.

[vmmedmed]

That's a big help. Thank you. It's such a strange experience consulting and having

such restricted access. But I can be more focused in my ask of the admins armed

with this.