There is not that much context here, so I will do a standard answer too.
If you have a doubt regarding the data owned by a node, running repair
takes some resources but should never break anything. I mean it is an
operation you can be running as much as you want. So I would use it, just
in case.
I have seen a bunch of them in the log files of some newly joined nodes. I did
a search in google and it seems increasing the countdown latch timeout can
solve this problem. But I assume it only resolves it for future nodes when
joining happens? For the existing nodes, anything needs to be done?
