From c330e77ec3b21f803f6e5926cefe166151ef4808 Mon Sep 17 00:00:00 2001 From: Allen Zhong Date: Wed, 12 May 2021 17:32:11 +0800 Subject: [PATCH] cluster: ignore error when increasing schedule limit during upgrade --- pkg/cluster/operation/upgrade.go | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/pkg/cluster/operation/upgrade.go b/pkg/cluster/operation/upgrade.go index ead4031266..51940e8cc9 100644 --- a/pkg/cluster/operation/upgrade.go +++ b/pkg/cluster/operation/upgrade.go @@ -64,7 +64,10 @@ func Upgrade( pdClient := api.NewPDClient(topo.(*spec.Specification).GetPDList(), 10*time.Second, tlsCfg) origLeaderScheduleLimit, origRegionScheduleLimit, err = increaseScheduleLimit(ctx, pdClient) if err != nil { - return err + // the config modifing error should be able to be safely ignored, as it will + // be processed with current settings anyway. + log.Warnf("failed increasing schedule limit: %s, ignore", err) + continue } defer func() { upgErr := decreaseScheduleLimit(pdClient, origLeaderScheduleLimit, origRegionScheduleLimit)