[jira] Updated: (CASSANDRA-1772) debian initscript sometimes mistakenly thinks it failed, gives extraneous output

["paul cannon (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/CASSANDRA-1772?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

paul cannon updated CASSANDRA-1772:
-----------------------------------

    Attachment: cass-add-status.patch.txt

cass-add-status.patch adds a slightly more robust is_running check (ensures the command line of the process $(cat $PIDFILE) matches what we expect, instead of just checking that there is a process with that number).

It also gets rid of the extraneous "ps" output to the terminal, and adds the "status" command to the initscript.

> debian initscript sometimes mistakenly thinks it failed, gives extraneous output
> --------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-1772
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1772
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Packaging
>    Affects Versions: 0.7.0 rc 1
>         Environment: Debian Squeeze with cassandra 0.7.0~rc1 on a slicehost VM
>            Reporter: paul cannon
>            Priority: Minor
>         Attachments: cass-add-status.patch.txt
>
>
> On my test systems, which are all relatively slow VMs, the Cassandra debian initscript usually thinks it fails to start, even though the startup was successful.  It appears that jsvc forks the daemon process and exits, and the initscript check for the running Cassandra service occurs before the new daemon is able to initialize itself and create its pidfile.
> On top of that, most invocations end up spitting out a small amount of garbage from /bin/ps, in addition to the typical "Stopping Cassandra: cassandra." log messages one sees if verbose=yes in /etc/default/rcS.  This is not very flattering.
> Finally, the initscript should provide the "status" command to meet current LSB spec. The functionality is mostly complete already anyway, and it can be quite useful.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.