Apache9 commented on code in PR #5865:
URL: https://github.com/apache/hbase/pull/5865#discussion_r1604729871
##########
hbase-server/src/main/java/org/apache/hadoop/hbase/master/replication/ReplicationPeerManager.java:
##########
@@ -402,6 +407,57 @@ public void removeAllQueuesAndHFileRefs(String peerId)
throws ReplicationExcepti
queueStorage.removePeerFromHFileRefs(peerId);
}
+ private void checkClusterKey(String clusterKey, ReplicationEndpoint endpoint)
+ throws DoNotRetryIOException {
+ if (endpoint != null && !(endpoint instanceof HBaseReplicationEndpoint)) {
+ return;
+ }
+ // Endpoints implementing HBaseReplicationEndpoint need to check cluster
key
+ URI connectionUri =
ConnectionRegistryFactory.tryParseAsConnectionURI(clusterKey);
+ try {
+ if (connectionUri != null) {
+ ConnectionRegistryFactory.validate(connectionUri);
+ } else {
+ ZKConfig.validateClusterKey(clusterKey);
+ }
+ } catch (IOException e) {
+ throw new DoNotRetryIOException("Invalid cluster key: " + clusterKey, e);
+ }
+ if (endpoint != null && endpoint.canReplicateToSameCluster()) {
+ return;
+ }
+ // make sure we do not replicate to same cluster
+ String peerClusterId;
+ try {
+ if (connectionUri != null) {
+ // fetch cluster id through standard admin API
+ try (Connection conn =
ConnectionFactory.createConnection(connectionUri, conf);
+ Admin admin = conn.getAdmin()) {
+ peerClusterId =
+
admin.getClusterMetrics(EnumSet.of(ClusterMetrics.Option.CLUSTER_ID)).getClusterId();
Review Comment:
I was also thinking about this when implementing, as well as get the region
server list, until I saw that in AsyncAdmin we have a getClusterId method but
the default implementation is to extract it from the cluster metrics...
Can file a new issue for implementing these methods directly.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
