wchevreuil commented on a change in pull request #494: HBASE-22380 break circle
replication when doing bulkload
URL: https://github.com/apache/hbase/pull/494#discussion_r325216175
##########
File path:
hbase-server/src/main/java/org/apache/hadoop/hbase/replication/regionserver/ReplicationSink.java
##########
@@ -243,14 +250,26 @@ public void replicateEntries(List<WALEntry> entries,
final CellScanner cells,
LOG.debug("Finished replicating mutations.");
}
- if (bulkLoadHFileMap != null && !bulkLoadHFileMap.isEmpty()) {
- LOG.debug("Started replicating bulk loaded data.");
- HFileReplicator hFileReplicator =
- new HFileReplicator(this.provider.getConf(this.conf,
replicationClusterId),
+ if(bulkLoadsPerClusters != null) {
+ for (List<String> clusterIds : bulkLoadsPerClusters.keySet()) {
Review comment:
That came up as well on the separate PR I created for master branch. Pasting
my explanation made there:
> HRegionServer already keeps a String clusterId (not UUID), which I'm using
to check against the incoming bulk load request. Converting the request List to
List when clusterId is already represented as String didn't seem to give much
gain here. Maybe we should raise a separate jira to convert all existing String
representations of cluster id into UUID, and including this one as part of this
work?
We agreed with that suggestion to open a separate jira to address replacing
existing String clusterId to UUID.
----------------------------------------------------------------
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
