Logical switch monitor ping fails but network works OK

This post was originally published on this site

Hi, I’m having problems with the Logical Switch Monitor feature that test VEPs communication. I have four LS in two compute clusters. Each cluster have its own DVS and the Transport Zone spans on both of them.
In three of the LSs I did the ping test under the LS Monitor area and they fail when I select either hosts from cluster-a as the source and either hosts of cluster-b as the destination.

One of the LS success in the same test, although all LS are created in the same manner. However, if I exchange the source and destination host for the ping test for that LS that works OK, it fails. Moreover, if I instead of doing the ping test I do the broadcast one, selecting as the source either hosts of cluster-a, it shows me that hosts from cluster-b are unresponsive, when the same host used as source host but using the ping test success!!!


I checked using the vmkping selecting the vxlan stack, size of 1572 and don’t fragment flag all the connections between al VEPs (every combinations of VTEPs between each other) and all of them were successful.
And lastly, I tested a ping from one VM in cluster-a to a VM in cluster-b, both connected to the LS that fails the ping test, using the don’t fragment flag and a size of 1472 (inside the VM OS the maximum size of a packet still remains at 1500 bytes) and it works OK!!!


So, my question is, could it be possible that the UI LS monitor ping and broadcast test do not work as one would expect? It is to say, can everything be working OK although the UI test fails? Or there is actually a problem I’m not seeing?




Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.