Fix clock skew issue

Signed-off-by: Florent Poinsard <[email protected]>

pkg/apis/planetscale/v2/vitessbackupschedule_methods.go

@@ -39,7 +39,7 @@ func (vbsc *VitessBackupSchedule) GetSuccessfulJobsLimit() int32 {
 // the VitessBackupSchedule will try catching up the missed scheduled runs one by one
 // this can be extremely lengthy in the even of a big clock skew, if the number of missed scheduled
 // jobs reaches either DefaultAllowedMissedRuns or the value specified by the user, the controller
-// will give up looking for the previously missed run and simply wait until the next scheduled run.
+// will give up looking for the previously missed run and error out.
 // Setting the default to 100 is fair, catching up a up to 100 missed scheduled runs is not lengthy.
 const DefaultAllowedMissedRuns = 100
 

pkg/controller/vitessbackupschedule/vitessbackupschedule_controller.go

@@ -295,10 +295,9 @@ func getNextSchedule(vbsc planetscalev2.VitessBackupSchedule, now time.Time) (ti
 		lastMissed = t
 		missedRuns++
 
-		// If we have too many missed jobs, just bail out as given on the clock lag, looping over the schedule might take forever.
+		// If we have too many missed jobs, just bail out as the clock lag is too big
 		if missedRuns > vbsc.GetMissedRunsLimit() {
-			log.Warnf("too many missed runs, skipping all previous runs and forwarding to the next scheduled time")
-			return time.Time{}, sched.Next(now), nil
+			return time.Time{}, time.Time{}, fmt.Errorf("too many missed runs, check clock skew or increase .spec.allowedMissedRun")
 		}
 	}