[
https://issues.apache.org/jira/browse/OAK-6387?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16169576#comment-16169576
]
Chetan Mehrotra commented on OAK-6387:
--------------------------------------
We can use the CheckpointMBean#getOldestCheckpointCreationDate (introduced with
OAK-6227) to determine the lower bound (max of that time and 24 hrs) for the
timestamp for binaries to keep. Any index files that are created and yet not
referred in NodeStore would be created after that.
[~catholicon] [~amitjain] Thoughts?
> Building an index (new index + reindex): temporarily store blob references
> --------------------------------------------------------------------------
>
> Key: OAK-6387
> URL: https://issues.apache.org/jira/browse/OAK-6387
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: lucene, query
> Reporter: Thomas Mueller
> Fix For: 1.8
>
>
> If reindexing a Lucene index takes multiple days, and if datastore garbage
> collection (DSGC) is run during that time, then DSGC may remove binaries of
> that index because they are not referenced.
> It would be good if all binaries that are needed, and that are older than
> (for example) one hour, are referenced during reindexing (for example in a
> temporary location). So that DSGC will not remove them.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
