siddhantsangwan commented on code in PR #3423:
URL: https://github.com/apache/ozone/pull/3423#discussion_r881473582
##########
hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/container/balancer/ContainerBalancer.java:
##########
@@ -801,24 +846,70 @@ private void resetState() {
/**
* Receives a notification for raft or safe mode related status changes.
* Stops ContainerBalancer if it's running and the current SCM becomes a
- * follower or enters safe mode.
+ * follower or enters safe mode. Starts ContainerBalancer if the current
+ * SCM becomes leader, is out of safe mode and balancer should run.
*/
@Override
public void notifyStatusChanged() {
- if (!scmContext.isLeader() || scmContext.isInSafeMode()) {
- if (isBalancerRunning()) {
- stopBalancingThread();
+ lock.lock();
+ try {
+ if (!scmContext.isLeader() || scmContext.isInSafeMode()) {
+ if (isBalancerRunning()) {
Review Comment:
In addition to my previous reply, I think holding state in ServiceStatus
along with persisting it in RocksDB would make the logic a bit complex. We'd
then have three ways of checking state - ServiceStatus, RocksDB, and checking
the current thread for null. What do you think @JacksonYao287 ?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
