[
https://issues.apache.org/jira/browse/SAMZA-375?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14098788#comment-14098788
]
Jon Bringhurst commented on SAMZA-375:
--------------------------------------
{quote}
Do we ever run into a race condition where the Mesos container has been
allocated and started, but the ZK state has not yet been saved?
{quote}
Right now I'm just following the pattern that I see in most Mesos examples
(which appears to have a race condition). Once things are working, it shouldn't
be much effort to do a prepare/committed and have the reregister handle the
state in between the two.
{quote}
Naive question: does Mesos provide any state facility itself? If the scheduler
fails and is restarted, will it tell you what was running? If so, does it give
you any metadata? Perhaps we could just store enough information in the
metadata about the already-existing containers.
{quote}
There's a few. When I say "stored in zookeeper", I really mean Mesos'
abstraction for ZK (org.apache.mesos.state.ZooKeeperState):
https://github.com/apache/mesos/tree/master/src/java/src/org/apache/mesos/state
I'll dig into SAMZA-348. It seems like an implementation of
org.apache.mesos.state.State that uses ConfigLog might be a possibility.
> Investigate Mesos Job Support
> -----------------------------
>
> Key: SAMZA-375
> URL: https://issues.apache.org/jira/browse/SAMZA-375
> Project: Samza
> Issue Type: Bug
> Reporter: Jon Bringhurst
> Assignee: Jon Bringhurst
>
> It would be nice if Samza had support for Mesos (https://mesos.apache.org/).
> The current plan is to create a MesosJob and MesosJobFactory, then look into
> what it would take to allow the AM code to act as a Mesos scheduler.
> The feasibility of this landing in trunk will be better understood after a
> rough prototype has been created.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
