[
https://issues.apache.org/jira/browse/MESOS-7181?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15939240#comment-15939240
]
Joseph Wu commented on MESOS-7181:
----------------------------------
An {{ExitedEvent}} is generated whenever a {{link(PID)}} breaks. This has dual
semantics depending on what the {{PID}} is:
* For local {{PID}} s, calling {{process::terminate(PID)}} will generate the
appropriate {{ExitedEvents}}.
* For remote {{PID}} s, we create a persistent socket connection. On breaking
the socket, we check a mapping of socket -> listeners and send {{ExitedEvents}}.
Semantically, what I'm proposing moving the semantics remote {{PID}} s closer
to that of local {{PID}} s. We would still keep all the existing behavior. We
would add an additional libprocess-level protocol in case a normal
{{MessageEvent}} is dropped.
I'm thinking of this approach:
* In {{ProcessManager::deliver}}, instead of logging {{VLOG(2) << "Dropping
event for process " << to;}}, we can respond with a special {{MessageEvent}} of
name {{TargetPIDExited}} (or something).
* In {{ProcessBase}}, we can install a {{MessageEvent}} handler for
{{TargetPIDExited}} and call {{ProcessBase::exited}} from there.
* To prevent infinity ping-pong-ing of these messages (in case both sender and
receiver actors terminated while their messages were in flight), the
{{process_manager}} can make an exception for this type of message.
> Stale frameworks seen on Mesos, but not known to scheduler
> ----------------------------------------------------------
>
> Key: MESOS-7181
> URL: https://issues.apache.org/jira/browse/MESOS-7181
> Project: Mesos
> Issue Type: Bug
> Components: general
> Reporter: Anindya Sinha
> Assignee: Anindya Sinha
>
> Using a scheduler which launches multiple frameworks using scheduler driver,
> we observe occasionally that a framework exists on Mesos which is not known
> to the scheduler. Since there is no entity that acts on the offers, this
> framework ends up hogging all the offers leading to starvation in the cluster.
> This particular scenario is as follows:
> 1) Scheduler does a driver.start() which results in the 1st SUBSCRIBE sent to
> master.
> 2) The scheduler driver resends the SUBSCRIBE (since the framework has not
> yet registered) which is a result of the exponential backoff.
> 3) Framework is registered based on the 1st SUBSCRIBE, but the scheduler
> issues a driver.stop() immediately which results in a TEARDOWN sent to the
> master.
> 4) Master processes the TEARDOWN which removes the framework.
> 5) Master now processes the 2nd SUBSCRIBE (after authorization) and tries to
> add this framework. This succeeds and a new framework id is generated (since
> the original framework is no longer registered after the TEARDOWN) but the
> Scheduler driver by now has already terminated once the scheduler issued the
> driver.stop(). So, master continues to send offers to this 2nd framework and
> hogs on to offers till offer time out.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)