[
https://issues.apache.org/jira/browse/PIG-205?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12590539#action_12590539
]
Alan Gates commented on PIG-205:
--------------------------------
At a glance, the patch looks fine. I think this raises a more fundamental
question though. What do we pass on the command line, and what do we expect to
have in the configuration file? I do not have a strong opinion about whether
the namenode goes on the command line or not. But I would like to have a
general policy to guide us. Currently we have the following items as command
line options that are also in the configuration file: log4jconf, cluster, hod
params, and exectype, and Mathieu proposes to add namenode. I can see three
options
1) Remove all these and only set them via config file
2) Determine a set that are important enough they should be easy to set via the
command line, everything else can only be set via the config file. I'm
guessing this set would be cluster, exectype, maybe namenode. Of course you'd
still be able to set these via config file.
3) Free for all, give command line options for whatever people want.
I definitely don't like 3, and 1 is probably not the best either. So I'd vote
for two.
Thoughts, discussion?
> Add a -n / -namenode parameter to pig command line
> --------------------------------------------------
>
> Key: PIG-205
> URL: https://issues.apache.org/jira/browse/PIG-205
> Project: Pig
> Issue Type: Improvement
> Reporter: Mathieu Poumeyrol
> Attachments: NameNodeArg.patch, NameNodeArg.v2.patch
>
>
> -c allows to specify the cluster job tracker location from the command line.
> For this to be usefull in most case, I expect users to need specifying the
> dfs location too.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
