[
https://issues.apache.org/jira/browse/MESOS-681?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Benjamin Mahler updated MESOS-681:
----------------------------------
Description:
Now that we have a reconciliation mechanism, we should document why it exists
and how to use it in 0.19.0 and going forward.
As we add the lower level API, reconciliation may be done slightly differently.
Having documentation that reflects the changes would be great.
It might also be helpful to upload my slides from the May 19th meetup.
was:
Once we have a Registrar in the master persistently storing slave information,
there will remain a smaller number of scenarios where updates can be lost in
the system. This ticket is to carefully think through and document these cases
so that we can identify possible improvements.
For example, if a slave machine fails and is not brought back up, and the
master fails over during this time. Then the master will send a slaveLost
signal to the scheduler (but will not be able to send TASK_LOST for each
individual task that was on the slave).
This way, framework developers will get a better sense of how reliable status
updates are, and under what circumstances they may not be delivered.
Summary: Document the 0.19.0 reconciliation API. (was: Document known
cases where updates can be lost.)
> Document the 0.19.0 reconciliation API.
> ---------------------------------------
>
> Key: MESOS-681
> URL: https://issues.apache.org/jira/browse/MESOS-681
> Project: Mesos
> Issue Type: Task
> Components: documentation
> Reporter: Benjamin Mahler
> Fix For: 0.19.0
>
>
> Now that we have a reconciliation mechanism, we should document why it exists
> and how to use it in 0.19.0 and going forward.
> As we add the lower level API, reconciliation may be done slightly
> differently. Having documentation that reflects the changes would be great.
> It might also be helpful to upload my slides from the May 19th meetup.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
