Starwind service hangs

[atlhivemind]

I have a 2 node cluster (just upgraded from v6 to v8).
Server 2012R2
I am attempting to shut down one node for maintenance/upgrade, so I stopped the Starwind VSAN Service and wait. On my other cluster node the process took about 10-15 minutes.

2 hours+ later:
- zero network activity
- no console access
- partner node shows this one is "offline"
- ram usage still high
- cpu usage and raid array usage has flatlined.

The last time this happened on the same node was under v6 during the upgrade. We (with the blessing of starwind support) force-killed the process. This took care of the immediate issue of starwind being stuck in a limbo state, but forced a full resync after the node came up. And 100+TB does not sync in less than a week...

Obviously that isn't a desirable outcome.

What gives?

[anton (staff)]

I'll bring engineers to take a look ASAP. Please make sure you do have StarWind logs "under hangs".

Back to sync: we're working on a LOG subsystem so even for FLAT you're not going to see full resync again (unless there's no data on the powered up node). LSFS is immune to that, we'll make FLAT immune too.