You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-issues@jackrabbit.apache.org by "Marcel Reutegger (JIRA)" <ji...@apache.org> on 2014/12/11 15:39:13 UTC
[jira] [Comment Edited] (OAK-2345) Diff reads too many nodes
[ https://issues.apache.org/jira/browse/OAK-2345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14242608#comment-14242608 ]
Marcel Reutegger edited comment on OAK-2345 at 12/11/14 2:39 PM:
-----------------------------------------------------------------
Fixed the background read to only consider external changes on a _lastRev if it is different from the previously seen value or the revision timestamp is within the time span the RevisionComparator maintains the seenAt revision.
Fixed in trunk: http://svn.apache.org/r1644650 and http://svn.apache.org/r1644654
was (Author: mreutegg):
Fixed the background read to only consider external changes on a _lastRev if it is different from the previously seen value or the revision timestamp is within the time span the RevisionComparator maintains the seenAt revision.
Fixed in trunk: http://svn.apache.org/r1644650
> Diff reads too many nodes
> -------------------------
>
> Key: OAK-2345
> URL: https://issues.apache.org/jira/browse/OAK-2345
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: core, mongomk
> Affects Versions: 1.0.8, 1.1.2, 1.1.3
> Reporter: Marcel Reutegger
> Assignee: Marcel Reutegger
> Fix For: 1.1.4
>
>
> DocumentNodeStore.diffManyChildren() may read too many nodes when there is an inactive cluster node with an old _lastRev on the root document. This is a regression introduced with the fix for OAK-2232.
> The fix assumes an inactive cluster node does not have a revision range with an old revision seen at a current timestamp. The DocumentNodeStore will in fact purge revisions from the range in the RevisionComparator after an hour. But on startup the first background read may populate the RevisionComparator with a revision, which is potentially very old (e.g. if the clusterId is not used anymore).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)