You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mesos.apache.org by "Chun-Hung Hsiao (JIRA)" <ji...@apache.org> on 2019/04/10 17:18:00 UTC
[jira] [Commented] (MESOS-8745) Expose inactive resource providers
through the `GET_RESOURCE_PROVIDERS` agent call.
[ https://issues.apache.org/jira/browse/MESOS-8745?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16814678#comment-16814678 ]
Chun-Hung Hsiao commented on MESOS-8745:
----------------------------------------
Some more details:
The “statuses” I’m thinking is something like `STAGED`, `STARTED` and `STOPPED`, to indicate that a RP config has been loaded, an RP instance has been instantiated, and it has been stopped (due to failures) w/ some message). If an RP is registered, we could omit its status from the list instead of adding some `RUNNING` status, because the sources of truth asynchronously come from two places (LRP daemon and RP manager) and might not be synced.
> Expose inactive resource providers through the `GET_RESOURCE_PROVIDERS` agent call.
> -----------------------------------------------------------------------------------
>
> Key: MESOS-8745
> URL: https://issues.apache.org/jira/browse/MESOS-8745
> Project: Mesos
> Issue Type: Improvement
> Components: agent
> Reporter: Chun-Hung Hsiao
> Priority: Major
> Labels: mesosphere, storage
>
> Once a resource provider config is placed through the {{ADD_RESOURCE_PROVIDER_CONFIG}} or {{UPDATE_RESOURCE_PROVIDER_CONFIG}} call, there is no way to tell if the resource provider fails to start. This makes it very hard for users to debug misconfigured resource providers, as they will need to dig into the agent log to figure out the error. We should report resource providers that are not registered yet in the {{GET_RESOURCE_PROVIDERS}} call, with the current statuses and error messages, to surface any error that fails a resource provider.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)