[
https://issues.apache.org/jira/browse/ZOOKEEPER-2141?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15217692#comment-15217692
]
Adam Milne-Smith commented on ZOOKEEPER-2141:
---------------------------------------------
Responses to each (now that I'm at a computer):
1. Absolutely - will provide a patch for this very shortly.
2. Good question - I would have assumed if a node were assigned no acls at all
then it should have no access rather than open unsafe. I believe there is no
change in behaviour from this patch though.
3. I'll correct this in the aforementioned patch.
4. It was to be able to lean on the Map.equals() where there is an AtomicLong
as the value - happy to make this change if desired.
5. I'll correct this in the aforementioned patch.
6. This is the intended behaviour as the line after the for loop calls it an
extra time and returns.
> ACL cache in DataTree never removes entries
> -------------------------------------------
>
> Key: ZOOKEEPER-2141
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2141
> Project: ZooKeeper
> Issue Type: Bug
> Affects Versions: 3.4.6
> Reporter: Karol Dudzinski
> Assignee: Adam Milne-Smith
> Priority: Blocker
> Fix For: 3.4.9, 3.5.2
>
> Attachments: ZOOKEEPER-2141-3.4.patch, ZOOKEEPER-2141.patch,
> ZOOKEEPER-2141.patch, ZOOKEEPER-2141.patch, ZOOKEEPER-2141.patch,
> ZOOKEEPER-2141.patch
>
>
> The problem and potential solutions are discussed in
> http://mail-archives.apache.org/mod_mbox/zookeeper-user/201502.mbox/browser
> I will attach a proposed patch in due course.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
