[
https://issues.apache.org/jira/browse/OAK-3804?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15063765#comment-15063765
]
Alex Parvulescu commented on OAK-3804:
--------------------------------------
bq. Optimally it would just print a warning and continue with what is there.
interesting point, I'll wrap the segment bits into a try/catch. not sure what
to do with the root candidates though, if an error is found, it should probably
discard the newer roots, as they are probably going to have the same SNFE
related issues..
For now I think it's best to go with the simple route: print all options out
and let someone decide what to do.
I think this could work quite well with the existing {{check}} [0], you
identify a candidate, set the head and run the consistency check.
[0] https://github.com/apache/jackrabbit-oak/tree/trunk/oak-run#check
> Add tarmk revision recovery listing to oak-run
> -----------------------------------------------
>
> Key: OAK-3804
> URL: https://issues.apache.org/jira/browse/OAK-3804
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: run, segmentmk
> Reporter: Alex Parvulescu
> Assignee: Alex Parvulescu
> Attachments: OAK-3804.patch
>
>
> There are cases where the flush thread apparently gets stuck (OAK-3303), this
> means that the journal file will not be updated resulting in loss of content
> as the journal file is the single source of truth.
> This oak run mode would be a best effort approach to parse the tar files and
> come out with (newer) candidates for root entries to the journal.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
