This message was deleted.

I couldn't find the associated github issue, can you link to it?

Here's the link: https://github.com/harvester/harvester/issues/3347

Having trouble reproducing this issue

As I mentioned on the GitHub page for this issue, changing the settings below in Longhorn solved the rebuild issues for us.

Concurrent Replica Rebuild Per Node Limit: 1 (value decreased)
Guaranteed Engine Manager CPU: 20 (value increased)
Guaranteed Replica Manager CPU: 20 (value increased)

Maybe the rebuild issues arise when the host machine is under heavy load by the rebuild process or the set CPU limits are not enough for the rebuild process to be performed properly. Just my thoughts on this. Maybe I have it completely wrong.