[jira] [Commented] (VCL-849) makeproduction state not processed correctly for cluster requests

["ASF subversion and git services (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/VCL-849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14391302#comment-14391302 ] 

ASF subversion and git services commented on VCL-849:
-----------------------------------------------------

Commit 1670760 from [~arkurth] in branch 'vcl/trunk'
[ https://svn.apache.org/r1670760 ]

VCL-849
Updated utils.pm::is_request_deleted to only return true if the request state or laststate is deleted. It had been returning true if either was makeproduction.pm.  Changed is_request_deleted to utilize get_request_current_state_name - the query was duplicated.

Added utils.pm::set_production_imagerevision. Moved this code from makeproduction.pm.

Fixed how makeproduction.pm::process sets log.ending. It had been setting it to 'EOR' regardless of the previous state. Changed so ending is set to 'deleted' if the previous state is new or reserved, and 'released' if inuse.

> makeproduction state not processed correctly for cluster requests
> -----------------------------------------------------------------
>
>                 Key: VCL-849
>                 URL: https://issues.apache.org/jira/browse/VCL-849
>             Project: VCL
>          Issue Type: Bug
>          Components: vcld (backend)
>            Reporter: Andy Kurth
>            Assignee: Andy Kurth
>            Priority: Minor
>             Fix For: 2.4.2
>
>
> The backend code does not handle the _makeproduction_ state properly for cluster requests.  The underlying problem lies in {{utils.pm::is_request_deleted}}.  It is returning true if the state is _deleted_ *or* _makeproduction_.  This throws off the code which waits for other cluster reservations to begin processing.  As a result, the process created for the _makeproduction_ state exits and vcld perpetually tries to fork another process.
> Josh is updating the frontend to prevent the _makeproduction_ option from showing for cluster requests.  Regardless, the backend code should be improved to make sure this doesn't happen for _makeproduction_ and other states.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)