falcon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Arpit Gupta (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FALCON-162) Update falcon scripts to let the user set the log and pid locations.
Date Fri, 08 Nov 2013 05:30:19 GMT

    [ https://issues.apache.org/jira/browse/FALCON-162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13817009#comment-13817009
] 

Arpit Gupta commented on FALCON-162:
------------------------------------

We populate the pid file after we start falcon so deleting the pid file before falcon starts
is not needed. Is there a particular reason why we dont want a pid file if a service is not
running? 

The only option is to remove the file as part of the stop script. We cannot handle the case
where if the user kills the process manually we will always have dangling pid file.

> Update falcon scripts to let the user set the log and pid locations.
> --------------------------------------------------------------------
>
>                 Key: FALCON-162
>                 URL: https://issues.apache.org/jira/browse/FALCON-162
>             Project: Falcon
>          Issue Type: Bug
>    Affects Versions: 0.3
>            Reporter: Arpit Gupta
>            Assignee: Arpit Gupta
>             Fix For: 0.4
>
>         Attachments: FALCON-162-1383346559.patch, FALCON-162-1383669144.patch, FALCON-162-1383750942.patch,
FALCON-162-1383789543.patch, FALCON-162-1383836907.patch, FALCON-162.patch, FALCON-162.patch,
FALCON-162.patch
>
>
> Current falcon scripts hard code the location of the log and the pid dirs. Instead if
we added the ability to source a falcon-env.sh which the user could provide then the user
can overwrite these locations to some thing else.
> This would also allow the user a way to set JAVA_HOME and other env vars in a file rather
than doing it in a shell. This is something similar to how hadoop works.
> Along with this change i also plan on changing the classpath to list dir/* instead of
every jar file. Again this is similar to what hadoop 2 does.



--
This message was sent by Atlassian JIRA
(v6.1#6144)

Mime
View raw message