Leader keeps sending messages to instance that was removed from cluster
Start cluster of 2 instances:
# terminal 1
picodata run --instance-id=i1 --cluster-id=c1 --listen=:3301 --peer=:3301 --data-dir=tmp-1
# terminal 2
picodata run --instance-id=i2 --cluster-id=c1 --listen=:3302 --peer=:3301 --data-dir=tmp-2
Stop second instance and remove all of it's data:
kill <i2 pid>
rm -r tmp-2
Restart i2:
picodata run --instance-id=i2 --cluster-id=c1 --listen=:3302 --peer=:3301 --data-dir=tmp-2
Now i2 has a different raft_id, and there's only 2 instances in the box.space.raft_group
, but i1
keeps sending messages to raft_id:2 which get ignored