huaxiangsun commented on a change in pull request #2522:
URL: https://github.com/apache/hbase/pull/2522#discussion_r502644247
##########
File path:
hbase-server/src/main/java/org/apache/hadoop/hbase/client/AsyncClusterConnection.java
##########
@@ -63,6 +64,13 @@
CompletableFuture<Long> replay(TableName tableName, byte[]
encodedRegionName, byte[] row,
List<Entry> entries, int replicaId, int numRetries, long
operationTimeoutNs);
+ /**
+ * Replicate wal edits for replica regions, the location for the replica is
already included.
+ * The return value is the edits we skipped, as the original return value is
useless.
+ */
+ CompletableFuture<Long> replay(TableName tableName, byte[]
encodedRegionName, List<Entry> entries,
+ int replicaId, int numRetries, long operationTimeoutNs, HRegionLocation
loc);
Review comment:
Yeah, thought about it. It is messy with my current implementation. The
whole idea is that since shipper only ships to one replica region sever and
passing in location will save extra cost to look up location.
For the case of region move case, the retry does not work in this
implementation as location is fixed. I think it will be better just let vanilla
implementation to take care of it. Let me clean this up and update the patch.
----------------------------------------------------------------
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]
