[
https://issues.apache.org/jira/browse/YARN-9033?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16723913#comment-16723913
]
Szilard Nemeth commented on YARN-9033:
--------------------------------------
Hi [~tangzhankun]!
I took an initial look on your description.
I guess you meant the following:
1. In {{NodeManager.serviceInit}}:
{{createResourcePluginManager}} gets called first, and then initialize is
called on the created object.
If the instance is {{LCE}}, {{ResourceHandlerChain.bootstrap()}} gets called
(as specified in {{LinuxContainerExecutor#init}}).
2. In {{NodeManager.serviceInit}} (later):
{{createContainerManager}} gets called and gets assigned to field
{{containerManager}}.
In the constructor of {{ContainerManagerImpl}}, a container scheduler is
created:
{{this.containerScheduler = createContainerScheduler(context);}}
3. At the last statement of {{NodeManager.serviceInit}}:
{{super.serviceInit(conf);}} gets called, that ultimately iterates over all the
services and calls {{serviceInit}} on them, including the {{LCE}} and the
{{ContainerScheduler}}.
I understand that we have 2 calls instead of 1 to
{{ResourceHandlerChain.bootstrap}} in case we use {{LCE}}.
I don't get the part what happens when we don't use {{LCE}}: Should we still
need to call bootstrap from the {{ContainerScheduler}}?
I didn't get this part specifically:
{quote}But actually, the "updateContainer" invocation in YARN-7715 depend on
containerId's cgroups path creation in "preStart" method which only happens
when we use "LinuxContainerExecutor".
{quote}
Where can I find this code part / what should I check?
Thanks!
> ResourceHandlerChain#bootstrap is invoked twice during NM start if
> LinuxContainerExecutor enabled
> -------------------------------------------------------------------------------------------------
>
> Key: YARN-9033
> URL: https://issues.apache.org/jira/browse/YARN-9033
> Project: Hadoop YARN
> Issue Type: Bug
> Components: yarn
> Reporter: Zhankun Tang
> Assignee: Zhankun Tang
> Priority: Major
> Attachments: YARN-9033-trunk.001.patch, YARN-9033-trunk.002.patch
>
>
> The ResourceHandlerChain#bootstrap will always be invoked in NM's
> ContainerScheduler#serviceInit (Involved by YARN-7715)
> So if LCE is enabled, the ResourceHandlerChain#bootstrap will be invoked
> first and then invoked again in ContainerScheduler#serviceInit.
> But actually, the "updateContainer" invocation in YARN-7715 depend on
> containerId's cgroups path creation in "preStart" method which only happens
> when we use "LinuxContainerExecutor". So the bootstrap of
> ResourceHandlerChain shouldn't happen in ContainerScheduler#serviceInit.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
