[ 
https://issues.apache.org/jira/browse/KNOX-2844?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17640497#comment-17640497
 ] 

ASF subversion and git services commented on KNOX-2844:
-------------------------------------------------------

Commit 616a4e06e8c24c6ba65250d4f853851d6a7d81b9 in knox's branch 
refs/heads/master from Attila Magyar
[ https://gitbox.apache.org/repos/asf?p=knox.git;h=616a4e06e ]

KNOX-2844 - Remote config monitor should not automatically delete local files 
if they're missing from the DB (#680)



> Remote config monitor should not automatically delete local files if they're 
> missing from the DB
> ------------------------------------------------------------------------------------------------
>
>                 Key: KNOX-2844
>                 URL: https://issues.apache.org/jira/browse/KNOX-2844
>             Project: Apache Knox
>          Issue Type: Sub-task
>            Reporter: Attila Magyar
>            Assignee: Attila Magyar
>            Priority: Major
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> Somehow we might need to differentiate between topologies which are 
> externally managed and topologies which are synced by the DB monitor. 
> Otherwise every hand deployed topology are getting deleted after enabling 
> this feature with an empty DB.
> By using logical deletes we can avoid unwanted local file removals. This 
> implementation is more similar than the ZK based implementation where the 
> delete event is generated by zookeeper, so if something never existed in ZK 
> it will never be deleted from the file system.
> The DB based implementation uses polling because we don't have these kind of 
> events (onDelete/onUpdate/onCreate). Originally the SQL based monitor removed 
> all the local files which were missing from the DB. But if we switch to 
> logical deletes, we can make sure that a file is only getting deleted if they 
> were previously added then (logically) removed from the DB.
> After a certain period, we can do a full cleanup on the table and do physical 
> removals on older, logiacally deleted records).



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to