You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Konstantin Shvachko (JIRA)" <ji...@apache.org> on 2007/03/23 03:14:32 UTC
[jira] Commented: (HADOOP-1085) Remove 'port rolling' from
Mini{DFS|MR}Cluster
[ https://issues.apache.org/jira/browse/HADOOP-1085?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12483378 ]
Konstantin Shvachko commented on HADOOP-1085:
---------------------------------------------
I am running "ant test-core" with this patch and without.
Here are my results
=============================================
Without 1085:
BUILD SUCCESSFUL
Total time: 19 minutes 2 seconds
=============================================
With 1085 patch:
[junit] Running org.apache.hadoop.dfs.TestReplication
[junit] Tests run: 1, Failures: 0, Errors: 1, Time elapsed: 29.525 sec
[junit] Test org.apache.hadoop.dfs.TestReplication FAILED
BUILD FAILED
/home/shv/kryptonite/hadoop/build.xml:461: Tests failed!
Total time: 19 minutes 18 seconds
=============================================
1) It breaks TestReplication
2) It runs slower.
3) Looking at the code I don't see how it eliminates dataNodeFirst parameter
There is a public construtor (not deprecated)
public MiniDFSCluster(int namenodePort,
Configuration conf,
int nDatanodes,
boolean dataNodeFirst,
boolean formatNamenode,
String[] racks) throws IOException {
and code inside supports the parameter.
if (dataNodeFirst) {
startDataNodes(conf, racks, data_dir);
}
4) I, like Owen, don't like sleeps. But this patch still uses them.
System.err.println("Waiting for the NameNode to initialize...");
Thread.sleep(1000);
> Remove 'port rolling' from Mini{DFS|MR}Cluster
> ----------------------------------------------
>
> Key: HADOOP-1085
> URL: https://issues.apache.org/jira/browse/HADOOP-1085
> Project: Hadoop
> Issue Type: Bug
> Components: test
> Affects Versions: 0.12.0
> Reporter: Arun C Murthy
> Assigned To: Arun C Murthy
> Fix For: 0.13.0
>
> Attachments: HADOOP-1085_20070311_1.patch
>
>
> The rolling of ports in these 2 clusters lead to a lot of timing issues and failed test cases; as witnessed in our patch process.
> The way around is to let the OS pick the port for the NameNode/JobTracker and let the the DataNode/TaskTracker query them for the port to connect to and then use that port.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.