I have tried nsx-manager 3.1 and have now back-leveled to 3.0.1 and my nsx-managers keep running out of memory and the manager goes down and GUI is not available. I am using a small form factor. The host is tight on memory as this is 16Gb required for the manager, but the host isn't running out of memory, the VM is. 3.0.1 isn't as bad, i can use it ... the manager goes down and then I have to wait maybe 5 mins for it to come back up as it seems to kill java processes and then restart the GUI. I can't see anything else wrong, it happens on all 4 of my esxi nodes (running latest vsphere 7) I keep getting sysrq mem on the console... with 3.1 I tried increasing memory to 20GB and it still runs out of memory. Any ideas on what I might be doing wrong? This is a lab and I can sort of live with it, but doesn't seem like this should be happening and I don't know where to start to diagnose it other than I am running top on the vm and see it continually eating more memory as I do any sort of work with the manager.
Is there any specific feature you have configured or just standard R&D features ? Does memory usage goes up without any operation or you executing some operations and then the usage goes high ?
Since this is lab you can try running "free -m" & "top" in privileged mode to verify the consumption in CLI.
I am just basically just deploying an edge appliance... I have had to back track and delete things occasionally, but I am in the very beginning stages of configuring NSX-T so I haven't really go much going on ... 3.0.1 seems to keep running consistently if I am not configuring anything on it. I have been running top with m and I see the memory use increase in it as I configure differently aspects - say to deploy my edge appliance. while it was deploying it ran out of memory polling to see when the appliance was ready. it seems ok at the moment, but this behaviour does not seem normal.
Is it a single manager or a cluster? I have seen high memory usage on small managers when clustered. I always recommend at least medium size manager to have better performance.
had something to do with the fact that my vsan cluster ESXi hosts did not have DNS setup. I neglected to set DNS after imaging them using VIA. I don't know how that affected nsx-manager memory, but somehow it did. After I realized I had missed a step in my setup procedure and configured the resolv.conf, my nsx-manager is using aroud 84% memory and working fine, not crashing anymore! Thanks for the input, I can't believe I managed to build a full vSAN cluster without something failing with no DNS configured 😞 I had been configuring my hosts by hand and then moved to using VIA and somehow had the impressions that I was feeding the DNS settings into it, but it does not, so problem solved.