[jira] [Commented] (OFBIZ-5626) No lock can be acquired after ofbiz application crashes.

["Jacques Le Roux (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/OFBIZ-5626?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14017615#comment-14017615 ] 

Jacques Le Roux commented on OFBIZ-5626:
----------------------------------------

Obviously it must be done just after the DB has been loaded and before any service or ECA fires. So I'd say at the end of GenericDAO.checkDb() Method. The one which finishes by <<timer.timerString("Finished Checking Entity Database");>>. Using something like entity.getEntityName() and plain SQL with JDBC. Just thinking aloud while reviewing code, but that should be it I guess.

I did a bit more review, this could be done between DatabaseUtil[268] and  DatabaseUtil[419]  using some code like in DatabaseUtil.createTable(). I must say I maybe did not digg enough...

> No lock can be acquired after ofbiz application crashes.
> --------------------------------------------------------
>
>                 Key: OFBIZ-5626
>                 URL: https://issues.apache.org/jira/browse/OFBIZ-5626
>             Project: OFBiz
>          Issue Type: Bug
>          Components: framework
>    Affects Versions: SVN trunk
>            Reporter: Leon
>            Priority: Critical
>
> We have a service which semaphore set to "fail". Someday, while it was running, the whole ofbiz crashed. After the restart, the service is unable to run anymore since the lock file (database record in ServiceSemaphore entity) is already there.
> Is there any way to cleanup the unused lock files during ofbiz startup? How about the cluster environment?



--
This message was sent by Atlassian JIRA
(v6.2#6252)