You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@hadoop.apache.org by Anfernee Xu <an...@gmail.com> on 2014/03/19 06:01:32 UTC

How to configure nodemanager.health-checker.script.path

Hello,

I'm running MR with 2.2.0 release, I noticed we can configure
"nodemanager.health-checker.script.path" in yarn-site.xml to customize NM
health checking, so I add below properties to yarn-site.xml

 <property>
     <name>yarn.nodemanager.health-checker.script.path</name>
     <value>/scratch/software/hadoop2/hadoop-dc/node_health.sh</value>
   </property>

  <property>
     <name>yarn.nodemanager.health-checker.interval-ms</name>
     <value>10000</value>
   </property>

To get a feel about this, the
/scratch/software/hadoop2/hadoop-dc/node_health.sh simply print an ERROR
message as below

#!/bin/bash
echo "ERROR disk full"
exit -1

But it seems not working, the node is still in health state, did I missing
something?

Thanks for your help.
-- 
--Anfernee

RE: How to configure nodemanager.health-checker.script.path

Posted by Rohith Sharma K S <ro...@huawei.com>.

Hi

Health script should execute successfully. If your health check required to fail, than add ERROR that print in console.  This is because health script may fail because of Syntax error, Command not found(IOexception) or several other reasons.

In order to work health script,
Do not add "exit -1".

#!/bin/bash
echo "ERROR disk full"

Thanks & Regards
Rohith Sharma K S

From: Anfernee Xu [mailto:anfernee.xu@gmail.com]
Sent: 19 March 2014 10:32
To: user
Subject: How to configure nodemanager.health-checker.script.path

Hello,

I'm running MR with 2.2.0 release, I noticed we can configure "nodemanager.health-checker.script.path" in yarn-site.xml to customize NM health checking, so I add below properties to yarn-site.xml

 <property>
     <name>yarn.nodemanager.health-checker.script.path</name>
     <value>/scratch/software/hadoop2/hadoop-dc/node_health.sh</value>
   </property>

  <property>
     <name>yarn.nodemanager.health-checker.interval-ms</name>
     <value>10000</value>
   </property>

To get a feel about this, the /scratch/software/hadoop2/hadoop-dc/node_health.sh simply print an ERROR message as below

#!/bin/bash
echo "ERROR disk full"
exit -1

But it seems not working, the node is still in health state, did I missing something?

Thanks for your help.
--
--Anfernee

RE: How to configure nodemanager.health-checker.script.path

Posted by Rohith Sharma K S <ro...@huawei.com>.

Hi

Health script should execute successfully. If your health check required to fail, than add ERROR that print in console.  This is because health script may fail because of Syntax error, Command not found(IOexception) or several other reasons.

In order to work health script,
Do not add "exit -1".

#!/bin/bash
echo "ERROR disk full"

Thanks & Regards
Rohith Sharma K S

From: Anfernee Xu [mailto:anfernee.xu@gmail.com]
Sent: 19 March 2014 10:32
To: user
Subject: How to configure nodemanager.health-checker.script.path

Hello,

I'm running MR with 2.2.0 release, I noticed we can configure "nodemanager.health-checker.script.path" in yarn-site.xml to customize NM health checking, so I add below properties to yarn-site.xml

 <property>
     <name>yarn.nodemanager.health-checker.script.path</name>
     <value>/scratch/software/hadoop2/hadoop-dc/node_health.sh</value>
   </property>

  <property>
     <name>yarn.nodemanager.health-checker.interval-ms</name>
     <value>10000</value>
   </property>

To get a feel about this, the /scratch/software/hadoop2/hadoop-dc/node_health.sh simply print an ERROR message as below

#!/bin/bash
echo "ERROR disk full"
exit -1

But it seems not working, the node is still in health state, did I missing something?

Thanks for your help.
--
--Anfernee

RE: How to configure nodemanager.health-checker.script.path

Posted by Rohith Sharma K S <ro...@huawei.com>.

Hi

Health script should execute successfully. If your health check required to fail, than add ERROR that print in console.  This is because health script may fail because of Syntax error, Command not found(IOexception) or several other reasons.

In order to work health script,
Do not add "exit -1".

#!/bin/bash
echo "ERROR disk full"

Thanks & Regards
Rohith Sharma K S

From: Anfernee Xu [mailto:anfernee.xu@gmail.com]
Sent: 19 March 2014 10:32
To: user
Subject: How to configure nodemanager.health-checker.script.path

Hello,

I'm running MR with 2.2.0 release, I noticed we can configure "nodemanager.health-checker.script.path" in yarn-site.xml to customize NM health checking, so I add below properties to yarn-site.xml

 <property>
     <name>yarn.nodemanager.health-checker.script.path</name>
     <value>/scratch/software/hadoop2/hadoop-dc/node_health.sh</value>
   </property>

  <property>
     <name>yarn.nodemanager.health-checker.interval-ms</name>
     <value>10000</value>
   </property>

To get a feel about this, the /scratch/software/hadoop2/hadoop-dc/node_health.sh simply print an ERROR message as below

#!/bin/bash
echo "ERROR disk full"
exit -1

But it seems not working, the node is still in health state, did I missing something?

Thanks for your help.
--
--Anfernee

RE: How to configure nodemanager.health-checker.script.path

Posted by Rohith Sharma K S <ro...@huawei.com>.

Hi

Health script should execute successfully. If your health check required to fail, than add ERROR that print in console.  This is because health script may fail because of Syntax error, Command not found(IOexception) or several other reasons.

In order to work health script,
Do not add "exit -1".

#!/bin/bash
echo "ERROR disk full"

Thanks & Regards
Rohith Sharma K S

From: Anfernee Xu [mailto:anfernee.xu@gmail.com]
Sent: 19 March 2014 10:32
To: user
Subject: How to configure nodemanager.health-checker.script.path

Hello,

I'm running MR with 2.2.0 release, I noticed we can configure "nodemanager.health-checker.script.path" in yarn-site.xml to customize NM health checking, so I add below properties to yarn-site.xml

 <property>
     <name>yarn.nodemanager.health-checker.script.path</name>
     <value>/scratch/software/hadoop2/hadoop-dc/node_health.sh</value>
   </property>

  <property>
     <name>yarn.nodemanager.health-checker.interval-ms</name>
     <value>10000</value>
   </property>

To get a feel about this, the /scratch/software/hadoop2/hadoop-dc/node_health.sh simply print an ERROR message as below

#!/bin/bash
echo "ERROR disk full"
exit -1

But it seems not working, the node is still in health state, did I missing something?

Thanks for your help.
--
--Anfernee