[
https://issues.apache.org/jira/browse/KAFKA-2587?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14944171#comment-14944171
]
Jun Rao commented on KAFKA-2587:
--------------------------------
As for seeing consistent view in ZK, I am not 100% sure about some of the
corner cases. Let me describe the scenarios and see if [~fpj] can help answer.
The way that we propagate the acl changes is the following.
1. Acl cli writes the new acl A for a resource R under /acls/R -> A and also
writes a sequential node under /acl_changes with R as the value to indicate
that the acl from resource R has changed.
2. The Authorizer subscribes to child changes to /acl_changes. When the watcher
is fired, it reads all new sequential nodes under /acl_changes. From those
sequential nodes, it knows what resources' acls have changed and it will read
the latest acl for R from /acls/R.
A a couple of questions on this in some of the corner cases. If Authorizer
loses the ZK connection (session is still valid) btw the time that it reads all
child nodes under /acl_changes and that it reads the value of the sequential
node, when the Authorizer reconnects to another ZK server and tries to re-read
the sequential node, is it possible that the sequential node doesn't exist yet
on the new ZK server? If so, will the child watcher be fired again when the
sequential node is propagated to the new ZK server?
> Transient test failure: `SimpleAclAuthorizerTest`
> -------------------------------------------------
>
> Key: KAFKA-2587
> URL: https://issues.apache.org/jira/browse/KAFKA-2587
> Project: Kafka
> Issue Type: Sub-task
> Reporter: Ismael Juma
> Assignee: Parth Brahmbhatt
> Fix For: 0.9.0.0
>
>
> I've seen `SimpleAclAuthorizerTest ` fail a couple of times since its recent
> introduction. Here's one such build:
> https://builds.apache.org/job/kafka-trunk-git-pr/576/console
> [~parth.brahmbhatt], can you please take a look and see if it's an easy fix?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
