[
https://issues.apache.org/jira/browse/YARN-5434?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16121406#comment-16121406
]
Sunil G commented on YARN-5434:
-------------------------------
[~djp] and [~rkanter]
A quick doubt. I was trying to test graceful decommissioning feature by adding
hostname to exclude file and firing below command. It seems timeout param is
not taken into effect. I am not very sure whether I am missing some params in
CLI. Please help in same.
{noformat}
HW12715:bin sunilgovindan$ ./yarn rmadmin -refreshNodes -g 60000 -client
2017-08-10 15:47:11,176 WARN util.NativeCodeLoader: Unable to load
native-hadoop library for your platform... using builtin-java classes where
applicable
2017-08-10 15:47:11,263 INFO client.RMProxy: Connecting to ResourceManager at
/127.0.0.1:25003
Graceful decommissioning completed in 0 seconds.
{noformat}
Logs are as follows:
{noformat}
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.NodesListManager: refreshNodes
excludesFile /tmp/nodemanager-excludes
2017-08-10 15:47:11,447 INFO org.apache.hadoop.util.HostsFileReader: Refreshing
hosts (include/exclude) list
2017-08-10 15:47:11,447 INFO org.apache.hadoop.util.HostsFileReader: Adding a
node "127.0.0.1" to the list of excluded hosts from /tmp/nodemanager-excludes
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.NodesListManager: hostsReader
include:{} exclude:{127.0.0.1}
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.NodesListManager: Gracefully
decommission node localhost:25006 with state RUNNING
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.RMAuditLogger: USER=sunilgovindan
IP=127.0.0.1 OPERATION=refreshNodes TARGET=AdminService
RESULT=SUCCESS
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNodeImpl: Put Node
localhost:25006 in DECOMMISSIONING.
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNodeImpl: Preserve
original total capability: <memory:8192, vCores:8>
2017-08-10 15:47:11,447 INFO
org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNodeImpl:
localhost:25006 Node Transitioned from RUNNING to DECOMMISSIONING
2017-08-10 15:47:11,494 INFO
org.apache.hadoop.yarn.server.resourcemanager.ResourceTrackerService:
DECOMMISSIONING localhost:25006 is ready to be decommissioned
2017-08-10 15:47:11,494 INFO
org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNodeImpl: Deactivating
Node localhost:25006 as it is now DECOMMISSIONED
2017-08-10 15:47:11,494 INFO
org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNodeImpl:
localhost:25006 Node Transitioned from DECOMMISSIONING to DECOMMISSIONED
2017-08-10 15:47:11,495 INFO
org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacityScheduler:
Removed node localhost:25006 clusterResource: <memory:0, vCores:0>
{noformat}
> Add -client|server argument for graceful decom
> ----------------------------------------------
>
> Key: YARN-5434
> URL: https://issues.apache.org/jira/browse/YARN-5434
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: graceful
> Affects Versions: 2.8.0
> Reporter: Robert Kanter
> Assignee: Robert Kanter
> Priority: Blocker
> Fix For: 2.8.0, 3.0.0-alpha1
>
> Attachments: YARN-5343.001.patch, YARN-5343.002.patch,
> YARN-5343.003.patch
>
>
> We should add {{-client|server}} argument to allow the user to specify if
> they want to use the client-side graceful decom tracking, or the server-side
> tracking (YARN-4676).
> Even though the server-side tracking won't go into 2.8, we should add the
> arguments to 2.8 for compatibility between 2.8 and 2.9, when it's added. In
> 2.8, using {{-server}} would just throw an Exception.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
