After a migration via the migration coordinator from NSX-V (6.4) to NSX-T(3.2) we ran into a couple of issues. The one i will talk about today is a random appearing QoS policy that does not seem to be active but is enabled on the VMs.
I spoke a colleague who also had seen this issue after an upgrade from NSX-T so maybe it’s more widespread as it seems.
Note; With a recent migration on version 220.127.116.11 we dit not see this policy, so probably it’s fixed.
After migration from V to T we ran soms iPERF between different VMs to validate that the bandwidth of NSX-T was sufficient enough for the workloads. During the tests we noticed that on some VMs we where nog able to get an higher bandwidth as 100Mbps Because this value seemed too specific to be random we dived in further.
We found out that there was a policy named QosProfile1 attatched to the segments with a bandwith limit of 100MBps. Strange thing is that the boxes doesn’t seem to be highlighted but as stated we did see a hard limit on the 100MBps.
After some digging into the logfiles we did see that the migration coordinator created this QoS policy because it “was not found” during migration. We where not able to determine if this setting was somewhere configured in NSX-V
We removed the QoS policy from al the segments. This solved the issue for most of the VMs and we dit get iPERF results from about 3/4 GBits, so the problem seems to be solved.
However there where still a few VMs with seems the have the gapped. To fix this issue you can ” reset the port to default in nsx manager mode ->networking->Segments->ports