[
https://issues.apache.org/jira/browse/BIGTOP-1125?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sean Mackrory updated BIGTOP-1125:
----------------------------------
Attachment: bigtop-1125-test.txt
0001-BIGTOP-1125.-Return-value-does-not-reflect-status-ch.patch
Good point, Mark, although the return value was going to be 0 with or without
my patch. I've attached a patch that uses 'return' instead of 'exit' more
consistently and returns values that better represent the status, and I've also
attached a transcript of my test sequence, which I have read through and
verified that the output was always what I thought it should be.
I have only made changes for the single-process use case. The function for
starting and stopping multiple daemons also returns 0 unconditionally, as I
recall there being a reason for that, so perhaps lets wait on adding any extra
complexity to that function. Any users of the multiple-process feature would
need to query the status to know which had failed, anyway.
> Return value does not reflect status checks
> -------------------------------------------
>
> Key: BIGTOP-1125
> URL: https://issues.apache.org/jira/browse/BIGTOP-1125
> Project: Bigtop
> Issue Type: Bug
> Affects Versions: 0.6.0
> Reporter: Sean Mackrory
> Assignee: Sean Mackrory
> Fix For: 0.8.0
>
> Attachments:
> 0001-BIGTOP-1125.-Return-value-does-not-reflect-status-ch.patch,
> 0001-BIGTOP-1125.-Return-value-does-not-reflect-status-ch.patch,
> bigtop-1125-test.txt
>
>
> The init script for HBase RegionServers that supports multiple processes is
> always returning 0. It should return 0 on success of non-zero on failure. In
> the case of running multiple processes, failure is defined as a failure
> There are already different constants defined in the file (e.g. different
> values for failures in all process, failures in some processes, failures in
> no processes... etc.), we just need to return them properly.
> Although this affects the 0.7.0 RC, I don't consider it a big enough deal to
> warrant -1'ing the RC, which I'm still testing.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
