[
https://issues.apache.org/jira/browse/MESOS-2340?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14558702#comment-14558702
]
Marco Massenzio commented on MESOS-2340:
----------------------------------------
eh eh don't look at me for native English speaker :)
(but I will happily review and fix what I can)
One thing that bothers me a bit, though, are possible race conditions between a
Leader going away and a new one being elected: if the new Leader gets elected,
and creates the new JSON znode in the {{discovery}} group, _before_ the old one
gets removed by ZK, any watcher may be confused upon finding *two* znodes there.
We should clearly document this situation, and we should also provide clear
guidelines that, in such situations, the Watcher should retry after a brief
interval - in other words, any condition under which in the {{discovery}}
subtree there is {{count(znodes) > 1}} should be considered an error.
> Publish JSON in ZK instead of serialized MasterInfo
> ---------------------------------------------------
>
> Key: MESOS-2340
> URL: https://issues.apache.org/jira/browse/MESOS-2340
> Project: Mesos
> Issue Type: Improvement
> Reporter: Zameer Manji
> Assignee: haosdent
>
> Currently to discover the master a client needs the ZK node location and
> access to the MasterInfo protobuf so it can deserialize the binary blob in
> the node.
> I think it would be nice to publish JSON (like Twitter's ServerSets) so
> clients are not tied to protobuf to do service discovery.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
