zuston commented on code in PR #53:
URL: https://github.com/apache/incubator-uniffle/pull/53#discussion_r937293659
##########
client-mr/src/main/java/org/apache/hadoop/mapreduce/task/reduce/RssRemoteMergeManagerImpl.java:
##########
@@ -144,8 +144,9 @@ public RssRemoteMergeManagerImpl(String appId,
TaskAttemptID reduceId, JobConf j
try {
remoteConf.setInt("dfs.replication", replication);
remoteConf.setInt("dfs.client.block.write.retries", retries); // origin=3
- this.remoteFS = ShuffleStorageUtils.getFileSystemForPath(new
Path(basePath), remoteConf);
- } catch (IOException e) {
+ HadoopAccessorProvider.init();
+ this.remoteFS = HadoopAccessorProvider.getFileSystem(new Path(basePath),
remoteConf);
Review Comment:
Yes.The spark/mr will manage the credentials of HDFS delegation token for
all executors, so there is no need to relogin in shuffle manager client of
uniffle.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
