why should we set max_mds to 1 when upgradeMDS ?

[hhstu]

I just want upgrade mds from ceph-v1.18 -> v1.18.2
Environment:

• Rook version (use rook version inside of a Rook Pod):
  v1.12.9 - > v1.14.*

func (c *Cluster) upgradeMDS() error {

	logger.Infof("upgrading MDS cluster for filesystem %q", c.fs.Name)

	// 1. set allow_standby_replay to false
	if err := cephclient.AllowStandbyReplay(c.context, c.clusterInfo, c.fs.Name, false); err != nil {
		return errors.Wrap(err, "failed to setting allow_standby_replay to false")
	}

	// Standby-replay daemons are stopped automatically.
	if err := cephclient.FailAllStandbyReplayMDS(c.context, c.clusterInfo, c.fs.Name); err != nil {
		return errors.Wrap(err, "failed to fail mds agent in up:standby-replay state")
	}

	// 2. set max_mds to 1
	logger.Debug("start setting active mds count to 1")
	if err := cephclient.SetNumMDSRanks(c.context, c.clusterInfo, c.fs.Name, 1); err != nil {
		return errors.Wrapf(err, "failed setting active mds count to %d", 1)
	}

	// 3. wait for ranks to be 0
	if err := cephclient.WaitForActiveRanks(c.context, c.clusterInfo, c.fs.Name, 1, false, fsWaitForActiveTimeout); err != nil {
		return errors.Wrap(err, "failed waiting for active ranks to be 1")
	}

	// 4. stop standby daemons
	daemonName, err := cephclient.GetMdsIdByRank(c.context, c.clusterInfo, c.fs.Name, 0)
	if err != nil {
		return errors.Wrap(err, "failed to get mds id from rank 0")
	}
	daemonNameTokens := strings.Split(daemonName, "-")
	daemonLetterID := daemonNameTokens[len(daemonNameTokens)-1]
	desiredDeployments := map[string]bool{
		fmt.Sprintf("%s-%s-%s", AppName, c.fs.Name, daemonLetterID): true,
	}
	logger.Debugf("stop mds other than %s", daemonName)
	err = c.scaleDownDeployments(1, 1, desiredDeployments, false)
	if err != nil {
		return errors.Wrap(err, "failed to scale down deployments during upgrade")
	}
	logger.Debugf("waiting for all standbys gone")
	if err := cephclient.WaitForNoStandbys(c.context, c.clusterInfo, 120*time.Second); err != nil {
		return errors.Wrap(err, "failed to wait for stopping all standbys")
	}

	// 5. upgrade current active deployment and wait for it come back
	_, err = c.startDeployment(c.clusterInfo.Context, daemonLetterID)
	if err != nil {
		return errors.Wrapf(err, "failed to upgrade mds %q", daemonName)
	}
	logger.Debugf("successfully started daemon %q", daemonName)

	// 6. all other MDS daemons will be updated and restarted by main MDS code path

	// 7. max_mds & allow_standby_replay will be reset in deferred function finishedWithDaemonUpgrade

	return nil
}

[hhstu]

I have review the code, can we skip the func of upgradeMDS when a minor version ceph update ?