Deploy custom service on non hadoop node with Apache Ambari

1

 

I want to deploy a custom service onto non hadoop nodes using Apache Ambari. I have created a custom service inside /var/lib/ambari-server/resources/common-services as opposed to Hadoop's folder of: /var/lib/ambari-server/resources/stacks/HDP

And then I restarted my ambari-server. But with web, I cannot see my new Service.

Am I missing anything, do I have to register my custom service anywhere?

Hi, I don’t want to hardcode my service version into metainfo.xml, Can I do it?

<service>
  <name>DUMMY_APP</name>
  <displayName>My Dummy APP</displayName>
  <comment>This is a distributed app.</comment>
  <version>0.1</version> --------------This I don't want to hardcode, Can I doit?
  <components>
  ...
  </components>
</service>

hadoop service hortonworks-data-platform ambari

shareimprove this question

edited Aug 26 '15 at 12:48

 

 

asked Aug 17 '15 at 10:36

user1393608

53211 gold badge99 silver badges1818 bronze badges

•  
  I have 1 more question: How can I do hosts specific configuration using apache ambari? Can I do it or whatever configuration I have selected applies for all the hosts? – user1393608 Sep 1 '15 at 7:14 

add a comment

1 Answer

activeoldestvotes

0

 

You still need to specify in the stack definition that your service is available for that particular stack. Common Services is just a place to maintain a common set of service definitions which can be used as part of a stack definition through extension.

For Example, lets say you have created custom service 'MYSERVICE' with a version identifier of '1.0' and want it to be provided for the HDP 2.2 stack.

• You would need to place your service definition at the following location:

  /var/lib/ambari-server/resources/common-services/MYSERVICE/1.0
  

  This above directory would contain your metainfo.xml file and your configuration and package folder. This will be the base definition of your service.

  Note: It is also important to note that the version you specify in /var/lib/ambari-server/resources/common-services/MYSERVICE/1.0/metainfo.xml must match the version number you have indicated in the path. For our example that would be <version>1.0</version>.

• You will then also need to add an additional metainfo.xml file to the HDP 2.2 stack that will provide this service.

  /var/lib/ambari-server/resources/stacks/HDP/2.2/services/MYSERVICE/metainfo.xml
  

  The contents of this file would be:

  <?xml version="1.0"?>
  <metainfo>
    <schemaVersion>2.0</schemaVersion>
    <services>
      <service>
        <name>MYSERVICE</name>
        <extends>common-services/MYSERVICE/1.0</extends>
      </service>
    </services>
  </metainfo>
  

shareimprove this answer

answered Aug 17 '15 at 20:04

cjackson

1,39711 gold badge1111 silver badges2424 bronze badges

•  
  Thanks Cjackson!!!, but then I will not be able to install my service on bare machine without hadoop? So hadoop stack version is compulsary? I just want my service to be deployed on any OS machine – user1393608 Aug 18 '15 at 5:42 
•  
  /var/lib/ambari-server/resources/stacks/HDP is not 'Hadoops' folder. It's the Hortonworks Data Platform Stack. You could define your own stack and only list your custom service in that stack. – cjackson Aug 18 '15 at 13:05
•  
  Ok one more query, How can I add my custom service to web ui without having to go through installation of new cluster and adding other services? – user1393608 Aug 18 '15 at 13:33
•  
  Hi, I don’t want to hardcode my service version into metainfo.xml, Can I do it? <service> <name>DUMMY_APP</name> <displayName>My Dummy APP</displayName> <comment>This is a distributed app.</comment> <version>0.1</version> <components> ... </components> – user1393608 Aug 26 '15 at 12:45
•  
  Why don't you want to hardcode it? There has to be a value there. If you explain to me why you don't want to hardcode it and what you're trying to achieve I may be able to help further. – cjackson Aug 27 '15 at 15:34