You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mesos.apache.org by "Greg Mann (JIRA)" <ji...@apache.org> on 2018/02/01 06:29:00 UTC

[jira] [Commented] (MESOS-8399) Use unique ID for CSI plugin containers in SLRP.

    [ https://issues.apache.org/jira/browse/MESOS-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16348080#comment-16348080 ] 

Greg Mann commented on MESOS-8399:
----------------------------------

Short-term fix for tests:
{code}
commit 53de510ae846ce2eab707493d6ab833429b6cec7
Author: Chun-Hung Hsiao <ch...@mesosphere.io>
Date:   Wed Jan 31 18:34:16 2018 -0800

    Used unique CSI plugin names in storage local resource provider tests.

    To provide better test isolation, the SLRP tests should generate
    random unique names in CSI plugin configs, so that the plugin
    containers use different container IDs.

    Review: https://reviews.apache.org/r/65427/
{code}

> Use unique ID for CSI plugin containers in SLRP.
> ------------------------------------------------
>
>                 Key: MESOS-8399
>                 URL: https://issues.apache.org/jira/browse/MESOS-8399
>             Project: Mesos
>          Issue Type: Bug
>            Reporter: Chun-Hung Hsiao
>            Assignee: Chun-Hung Hsiao
>            Priority: Major
>              Labels: mesosphere, storage
>
> If an agent crashed abnormally and the runtime directory is lost, then a standalone container previously launched by SLRP will be considered orphan when the agent restarts. Since the orphan containers are cleaned up asynchronously, it is possible that the cleanup is racing with SLRP launching a new standalone container instance with the same ID. To avoid this race, we should use unique IDs for CSI plugin containers.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)