replication: set RequeueAfter to 30 secondary for requeue #123
Conversation
Setting Requeue time for 30 seconds as the resync can take time and having default Requeue exponential backoff time can affect the RTO time closes: csi-addons#120 Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
|
@Madhu-1 the referred issue says of requeing at 15seconds where this PR is on 30 seconds. so would be good to correct either of the place. Also, ooc, do we have any of the controller references which does 30 second reque? |
@humblec what controllers you are looking at here? the resync is something async operation and it does not fix the exponential backoff. |
| @@ -263,7 +263,13 @@ func (r *VolumeReplicationReconciler) Reconcile(ctx context.Context, req ctrl.Re | |||
| logger.Info("volume is not ready to use, requeuing for resync") | |||
|
|
|||
| _ = r.updateReplicationStatus(instance, logger, getCurrentReplicationState(instance), "volume is degraded") | |||
| return ctrl.Result{Requeue: true}, nil | |||
| return ctrl.Result{ | |||
There was a problem hiding this comment.
I would add the same to the replicationErr condition above, as during resync post initial errors it goes into an exponential backoff as well post the initial 15 second check at line 231?
IOW, a resync check moves to exponential backoff if it receives errors, this does not impact RTO as failover/recover is not impacted by waiting for the failed cluster to establish a resync.
Setting Requeue time for 30 seconds as the resync can take time and having default Requeue exponential backoff time can affect the RTO time
closes: #120
Signed-off-by: Madhu Rajanna madhupr007@gmail.com