symious commented on a change in pull request #1776:
URL: https://github.com/apache/ozone/pull/1776#discussion_r557833774
##########
File path:
hadoop-ozone/ozonefs-hadoop2/src/main/java/org/apache/hadoop/fs/ozone/Hadoop27RpcTransport.java
##########
@@ -67,6 +69,12 @@ public OMResponse submitRequest(OMRequest payload) throws
IOException {
try {
return proxy.submitRequest(NULL_RPC_CONTROLLER, payload);
} catch (ServiceException e) {
+ OMNotLeaderException notLeaderException =
Review comment:
@elek Thanks for the comment.
Can reproduce the error by the following steps:
1. Prepare an OM-HA cluster.
2. Prepare the config files in HADOOP_CONF_DIR, including the core-site.xml
and the ozone-site.xml.
3. Run the command of "hdfs dfs -ls o3fs://bucket.volume.cluster1/" on a
follower OM. The client would get the NotLeaderExecption.
4. Run the command of "hdfs dfs -ls o3fs://bucket.volume.cluster1/" on the
leader OM. The client would get normal result.
5. Run the command of "hdfs dfs -ls o3fs://bucket.volume.cluster1/" on other
servers. The client would be retring to connect to "0.0.0.0:9862".
When Hadoop27RpcTransport is trying to initialize the socket to OM, client
would use the config of "ozone.om.address", and for an OM-HA cluster, this
value should be default as "0.0.0.0".
So when client initialize the proxy on a follower OM host, it will try to
connect to the follower OM, and an NotLeaderException will be thrown. And if
the client is on the leader server, the proxy would be successfully created and
no exceptions should be thrown. And If the client is on other servers, it will
try to connect to the local OM, while on OM is running, so the client would be
retrying to connect "0.0.0.0:9862".
----------------------------------------------------------------
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
