[
https://issues.apache.org/jira/browse/MESOS-143?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13249060#comment-13249060
]
[email protected] commented on MESOS-143:
-----------------------------------------------------
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/4533/
-----------------------------------------------------------
(Updated 2012-04-06 23:44:54.293983)
Review request for mesos and Benjamin Hindman.
Summary
-------
When I started this project, it was somewhat complicated, but with the recent
update that passes FrameworkInfo protobufs into the MesosSchedulerDriver, all
that was necessary to allow frameworks to set their own timeouts was to add
timeout as a field in FrameworkInfo and tell master.cpp to use this value. I
made it optional since some people might not care to set it and it would just
be an added complication, but I also added it to all three of the
test_frameworks (C++, Java, Python) to demonstrate its use.
The choice for DEFAULT_FRAMEWORK_FAILOVER_TIMEOUT and
MAX_FRAMEWORK_FAILOVER_TIMEOUT was basically arbitrary, so I'm open to
suggestions on what might be more appropriate.
This addresses bug MESOS-143.
https://issues.apache.org/jira/browse/MESOS-143
Diffs (updated)
-----
src/master/master.cpp 4dc9ee0
src/master/master.hpp 8a34d7e
src/master/constants.hpp 8248475
include/mesos/mesos.proto 75eee57
include/mesos/scheduler.hpp c2acf0d
Diff: https://reviews.apache.org/r/4533/diff
Testing
-------
Thanks,
Thomas
> Framework failover timeout should be specified by the framework.
> ----------------------------------------------------------------
>
> Key: MESOS-143
> URL: https://issues.apache.org/jira/browse/MESOS-143
> Project: Mesos
> Issue Type: Improvement
> Reporter: Benjamin Hindman
> Assignee: Thomas Marshall
> Priority: Blocker
>
> The framework failover timeout should be specified by the framework, with
> perhaps an upper limit enforced by the master. Right now two frameworks that
> want different failover timeout semantics is pretty much impossible (default
> right now is 1 second, favoring more "batch" like frameworks).
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
