Hi,
Lately we've been seeing some weird behavior when using the cluster upgrade feature in 8.10.x. We used to use this feature all the time without issues, but now we've had several end customer facing the same issue (again today going to 8.10.0.12):
Let's say they have about 300 AP's, the software gets uploaded to both MD's (from the MC). MD1 reboots and the AP's start failing over. All of a sudden 200+ AP's are upgrading at the same time, and after they are again active, the others are properly divided and upgrading as they are supposed to.
When checking "show lc-cluster <CLUSTER> upgrade status verbose", we can see that group/partition id 1 contains most AP's ( 200+) and all other groups/partition id(2-16) only contain about 5 AP's each...
Checking the 5GHz channel distribution, all seems ok. Also the LC-cluster is working as supposed. Anyone have any explanation why this is happening? Is there a command to calculate those part-id again to get a more evenly distributed grouping?
Thanks
Peter