[
https://issues.apache.org/jira/browse/OAK-4494?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15346228#comment-15346228
]
Vikas Saurabh commented on OAK-4494:
------------------------------------
Yup, sounds fine.
btw,
{code}
Long modCount = Utils.asLong((Number) obj.get(Document.MOD_COUNT));
- modCounts.put(id, modCount);
+ if (modCount == null) {
+ modCount = -1L;
+ }
+ Long modified = Utils.asLong((Number) obj.get(Document.MOD_COUNT));
+ if (modified == null) {
+ modified = -1L;
+ }
+ modCounts.put(id, new ModificationStamp(modCount, modified));
{code}
I think the second obj.get was supposed to be done on MODIFIED_IN_SECS.
I think the tests pass because duplicated modCount was different than cached
modCount (which, imo, is a good sign)... but, should we worry that the tests
didn't catch it?
> Stale documents after revision GC in cluster
> --------------------------------------------
>
> Key: OAK-4494
> URL: https://issues.apache.org/jira/browse/OAK-4494
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: core, documentmk, mongomk, rdbmk
> Reporter: Marcel Reutegger
> Assignee: Marcel Reutegger
> Priority: Minor
> Labels: candidate_oak_1_0, candidate_oak_1_2, candidate_oak_1_4
> Fix For: 1.6
>
> Attachments: OAK-4494.patch
>
>
> Implement additional tests for revision GC in a cluster.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
