[
https://issues.apache.org/jira/browse/YARN-8925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16690924#comment-16690924
]
Weiwei Yang commented on YARN-8925:
-----------------------------------
Hi [~Tao Yang]
I took sometime tested this patch with trunk build. I used configuration based
node-attribute provider.
Initially I set node-attributes as following
{code:java}
<property>
<name>yarn.nodemanager.node-attributes.provider.configured-node-attributes</name>
<value>osType,STRING,redhat:osVersion,STRING,2.6</value>
</property>
{code}
then I start RM/NM, from RM log I see
{noformat}
2018-11-18 14:44:37,365 INFO
org.apache.hadoop.yarn.server.resourcemanager.ResourceTrackerService:
NodeManager from node ip-172-31-15-90.us-east-2.compute.internal(cmPort: 37873
httpPort: 8042) registered with capability: <memory:8192, vCores:8>, assigned
nodeId ip-172-31-15-90.us-east-2.compute.internal:37873, node attributes \{
[nm.yarn.io/osType(STRING)=redhat, nm.yarn.io/osVersion(STRING)=2.6] }
{noformat}
it seems node-attributes are correctly registered.
then on NM logs, I kept seeing following error message
{noformat}
2018-11-18 14:44:37,443 ERROR
org.apache.hadoop.yarn.server.nodemanager.NodeStatusUpdaterImpl: NM node
attributes \{[nm.yarn.io/osType(STRING)=redhat,
nm.yarn.io/osVersion(STRING)=2.6]} were not accepted by RM and message from RM
: null
{noformat}
since there is no update, I suppose in this case, NM should report null to RM
and RM just simply ignore the message right?
Then I update the node-attributes by changing the value of the property
{code:java}
osVersion,STRING,2.7
{code}
then I expect to see this update should be reported back to RM and RM should
accept this change, however, I am still seeing the log
{noformat}
2018-11-18 14:56:36,771 ERROR
org.apache.hadoop.yarn.server.nodemanager.NodeStatusUpdaterImpl: NM node
attributes \{[nm.yarn.io/osType(STRING)=redhat,
nm.yarn.io/osVersion(STRING)=2.7]} were not accepted by RM and message from RM
: null
{noformat}
it doesn't look like to be working as expected, could you pls take a look?
Thanks
> Updating distributed node attributes only when necessary
> --------------------------------------------------------
>
> Key: YARN-8925
> URL: https://issues.apache.org/jira/browse/YARN-8925
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager
> Affects Versions: 3.2.1
> Reporter: Tao Yang
> Assignee: Tao Yang
> Priority: Major
> Labels: performance
> Attachments: YARN-8925.001.patch, YARN-8925.002.patch,
> YARN-8925.003.patch, YARN-8925.004.patch, YARN-8925.005.patch,
> YARN-8925.006.patch, YARN-8925.007.patch
>
>
> Currently if distributed node attributes exist, even though there is no
> change, updating for distributed node attributes will happen in every
> heartbeat between NM and RM. Updating process will hold
> NodeAttributesManagerImpl#writeLock and may have some influence in a large
> cluster. We have found nodes UI of a large cluster is opened slowly and most
> time it's waiting for the lock in NodeAttributesManagerImpl. I think this
> updating should be called only when necessary to enhance the performance of
> related process.
--
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
