You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mesos.apache.org by "Benjamin Bannier (JIRA)" <ji...@apache.org> on 2017/12/21 17:04:00 UTC
[jira] [Assigned] (MESOS-8349) When a resource provider driver is
disconnected, it fails to reconnect.
[ https://issues.apache.org/jira/browse/MESOS-8349?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Benjamin Bannier reassigned MESOS-8349:
---------------------------------------
Assignee: Benjamin Bannier (was: Jan Schlicht)
> When a resource provider driver is disconnected, it fails to reconnect.
> -----------------------------------------------------------------------
>
> Key: MESOS-8349
> URL: https://issues.apache.org/jira/browse/MESOS-8349
> Project: Mesos
> Issue Type: Bug
> Affects Versions: 1.5.0
> Reporter: Jan Schlicht
> Assignee: Benjamin Bannier
> Labels: mesosphere
>
> If the resource provider manager closes the HTTP connection of a resource provider, the resource provider should reconnect itself. For that, the resource provider driver will change its state to "DISCONNECTED", call a {{disconnected}} callback and use its endpoint detector to reconnect.
> This doesn't work in a testing environment where a {{ConstantEndpointDetector}} is used. While the resource provider is notified of the closed HTTP connection (and logs {{End-Of-File received}}), it never disconnects itself and calls the {{disconnected}} callback. Discarding {{HttpConnectionProcess::detection}} in {{HttpConnectionProcess::disconnected}} doesn't trigger the {{onAny}} callback of that future. This might not be a problem in {{HttpConnectionProcess}} but could be related to the test case using a {{ConstantEndpointDetector}}.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)