saintstack commented on a change in pull request #1071: HBASE-23693 Split
failure may cause region hole and data loss when use zk assign
URL: https://github.com/apache/hbase/pull/1071#discussion_r368651213
##########
File path:
hbase-server/src/main/java/org/apache/hadoop/hbase/master/RegionStates.java
##########
@@ -737,11 +741,13 @@ public void regionOffline(
public List<HRegionInfo> serverOffline(final ZooKeeperWatcher watcher, final
ServerName sn) {
// Offline all regions on this server not already in transition.
List<HRegionInfo> rits = new ArrayList<HRegionInfo>();
- Set<HRegionInfo> regionsToClean = new HashSet<HRegionInfo>();
+ Set<Pair<HRegionInfo, HRegionInfo>> regionsToClean =
+ new HashSet<Pair<HRegionInfo, HRegionInfo>>();
// Offline regions outside the loop and synchronized block to avoid
// ConcurrentModificationException and deadlock in case of meta anassigned,
// but RegionState a blocked.
Set<HRegionInfo> regionsToOffline = new HashSet<HRegionInfo>();
+ Map<String, HRegionInfo> daughter2Parent = new HashMap<>();
Review comment:
Suggest this needs more comment on why we need this accounting in
daughter2Parent.
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
