[jira] [Comment Edited] (OAK-8521) Add shutdown hook for oak-run commands

["Marcel Reutegger (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/OAK-8521?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16901016#comment-16901016 ] 

Marcel Reutegger edited comment on OAK-8521 at 8/6/19 1:03 PM:
---------------------------------------------------------------

The changes look good to me, except for the {{RevisionsCommand}}. This command already has a shutdown hook and none of the sub-commands constructs a read-write DocumentNodeStore. In this case there's no adverse effect when the DocumentNodeStore is not closed when the process ends. In fact, the proposed shutdown hook for the closer causes an exception when a QUIT signal is sent to a running {{revisions collect --continuous}} command.


was (Author: mreutegg):
The changes look good to me, except for the {{RevisionsCommand}}. This command already has a shutdown hook and none of the sub-commands constructs a read-write DocumentNodeStore. In this case there's no adverse effect when the DocumentNodeStore is not closed when the process ends. In fact, the proposed shutdown hook for the closer causes an exception when a QUIT signal is sent to a running {{revisions collect}} command.

> Add shutdown hook for oak-run commands
> --------------------------------------
>
>                 Key: OAK-8521
>                 URL: https://issues.apache.org/jira/browse/OAK-8521
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: documentmk, run
>            Reporter: Vinod Holani
>            Assignee: Marcel Reutegger
>            Priority: Minor
>         Attachments: OAK-8521.patch
>
>
> Ensure there are shutdown hooks in place for the various oak-run commands that properly stop the repository and release the clusterId.
> Many command construct a repository, then perform some operation and then shut down the repository again. Shutting down the repository will release the clusterId change the state of the clusterNodes entry back to inactive. In some cases commands execute for a longer period of time and a user might kill the process. When this happens, the clusterId is not released and stays in the active state with a lease that eventually expires. The command implementations should be improved to register a JVM shutdown hook that ensures the repository is shut down, even when the process is killed.



--
This message was sent by Atlassian JIRA
(v7.6.14#76016)