Connor
Connor
It turns out the io flow difference between raft-engine and kvdb. As raft-engine performs lz4 compression on write while kvdb doesn't enable wal compression, the size after compression would be...
It's better to handle the destroying peers in a separate thread pool like how generating snapshots does.
Please fix the conflict
> > Please fix the conflict > > Tried rebasing with master and found the working tree clean. Any advice? What do you mean by "working tree clean"
Regenerate the dashboard, then push
@LykxSassinator Probabaly. Anyway, we should avoid IO in raftstore thread
> Do we have any situations that a snapshot is reused? @Connor1996 Seems no
@zhangjinpeng87 We only added the metric, still need to figure out the cause of the high duration. Better keep this issue open
It's caused by an orphan peer.  From the log, the peer is applying a snapshot while told being stale and needs to be destroyed. Due to code defect, after...