[GitHub] [hbase] bharathv commented on a change in pull request #2899: HBASE-25523 Region normalizer chore thread is getting killed

[GitBox <gi...@apache.org>]

bharathv commented on a change in pull request #2899:
URL: https://github.com/apache/hbase/pull/2899#discussion_r562129209



##########
File path: hbase-server/src/main/java/org/apache/hadoop/hbase/master/normalizer/SimpleRegionNormalizer.java
##########
@@ -222,8 +223,12 @@ public void setMasterRpcServices(MasterRpcServices masterRpcServices) {
   private long getRegionSize(HRegionInfo hri) {
     ServerName sn = masterServices.getAssignmentManager().getRegionStates().
       getRegionServerOfRegion(hri);
-    RegionLoad regionLoad = masterServices.getServerManager().getLoad(sn).
-      getRegionsLoad().get(hri.getRegionName());
+    ServerLoad load = masterServices.getServerManager().getLoad(sn);
+    if (load == null) {
+      LOG.debug(hri.getRegionNameAsString() + " was not found on any server");

Review comment:
       Saw this issue today.. few comments
   
   Doesn't this mean the server "sn" is offline (not that region is not assigned)? 
   
   A tighter check here is isRegionOnline(), first check if the region is online (which ensures the region is not in transition and is assigned), we can even loop a few times to avoid any transient RITs.
   
   Another related question is does it make sense to run normalizer if there are RITs.




----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org