[BUG]koord-manager failed to renew lease about koordinator HOT 5 OPEN

@b43646 Do you mean the koord-manager restarts when the PodGroup is submitted?

from koordinator.

@saintube It seems that the pod restarted three days after the podgroup was submitted because the koord-manager failed to renew the lease.

from koordinator.

@saintube It seems that the pod restarted three days after the podgroup was submitted because the koord-manager failed to renew the lease.

@b43646 It might be a common issue of the cluster environment where the koord-manager queries the apiserver and timeout. Is there any additional clue to show the koord-manager works abnormally so we can investigate?

from koordinator.

@saintube The network is fluctuating and unstable. Does the koord-manager have a retry mechanism when renewing the lease?

from koordinator.

@saintube The network is fluctuating and unstable. Does the koord-manager have a retry mechanism when renewing the lease?

@b43646 Yes. The koord-manager uses the leader election mechanism of the controller-runtime, so you can find the manager is still working after the twice lease lost.

from koordinator.