You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by GitBox <gi...@apache.org> on 2020/11/10 03:05:47 UTC

[GitHub] [hbase] Apache9 commented on a change in pull request #2640: Backport: HBASE-25126 Add load balance logic in hbase-client to distribute read…

Apache9 commented on a change in pull request #2640:
URL: https://github.com/apache/hbase/pull/2640#discussion_r520255266



##########
File path: hbase-client/src/main/java/org/apache/hadoop/hbase/client/ConnectionImplementation.java
##########
@@ -161,7 +165,11 @@
   private final boolean hostnamesCanChange;
   private final long pause;
   private final long pauseForCQTBE;// pause for CallQueueTooBigException, if specified
-  private boolean useMetaReplicas;

Review comment:
       I can not recall whether I have mentioned this on the patch for master. The scope of 'useMetaReplicas' is wider than the newly introduced metaReplicaMode, as it could also effect the methods in Admin implementation, so I do not think we could remove it directly. If we want to do so, we'd better follow the typical rule to deprecated it first and then remove it in the next major release.

##########
File path: hbase-client/src/main/java/org/apache/hadoop/hbase/client/ReversedClientScanner.java
##########
@@ -64,7 +64,9 @@ protected boolean setNewStartKey() {
 
   @Override
   protected ReversedScannerCallable createScannerCallable() {
+    // In case of meta, we want to support scan over a replica region.
+    int replicaId = (scan.getConsistency() == Consistency.TIMELINE) ? scan.getReplicaId() : 0;

Review comment:
       This is a bit strange, TIMELINE means we should send request to primary first and if it does not come back in time, we switch to secondary replicas. IIRC the design here is that, if targetReplicaId is -1, we follow the rule of STRONG or TIMELINE, if targetReplicaId is not -1, we will go to the specific replica directly?

##########
File path: hbase-client/src/main/java/org/apache/hadoop/hbase/client/ScannerCallableWithReplicas.java
##########
@@ -42,13 +42,15 @@
 import org.slf4j.LoggerFactory;
 import org.apache.hadoop.hbase.client.ScannerCallable.MoreResults;
 import org.apache.hadoop.hbase.util.Pair;
+import static org.apache.hadoop.hbase.client.RegionReplicaUtil.DEFAULT_REPLICA_ID;

Review comment:
       Put static import on top?




----------------------------------------------------------------
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