Actually it seems that this command returns the same error

whirr.instance-templates=1 hadoop-namenode+hadoop-jobtracker, 2
hadoop-datanode+hadoop-tasktracker

but this one doesn't

whirr.instance-templates=1
hadoop-namenode+hadoop-jobtracker+hadoop-datanode+hadoop-tasktracker

It looks like it's breaking when there are multiple groupings of instances.


On Wed, Jan 4, 2012 at 3:41 PM, Andrei Savu <[email protected]> wrote:

> Let me know if a similar recipe for Hadoop works with Whirr 0.7.0 - it
> seems like the trunk is a bit broken now.
>
> -- Andrei Savu
>
>
> On Wed, Jan 4, 2012 at 10:30 PM, David Dundas <[email protected]> wrote:
>
>> Ok that recipe works. It also looks like something in my original
>>  whir.instance-templates
>>
>> May have been causing an error as well.
>>
>> Thanks!
>>
>> On Wed, Jan 4, 2012 at 10:48 AM, Andrei Savu <[email protected]>wrote:
>>
>>> I have done some testing and I have been able to start a fully
>>> functional ZooKeeper cluster from a VM also running inside the Amazon Cloud
>>> using a recipe like this:
>>>
>>> whirr.cluster-name=zookeeper-asavu-from-ec2
>>> whirr.instance-templates=3 zookeeper
>>> whirr.cluster-user=asavu
>>> whirr.provider=aws-ec2
>>> whirr.identity=${env:AWS_ACCESS_KEY_ID}
>>> whirr.credential=${env:AWS_SECRET_ACCESS_KEY}
>>>
>>> Hadoop fails with the following exception but this is unrelated to this
>>> discussion:
>>>
>>> Authorizing firewall ingress to [us-east-1/i-e2187780] on ports [8020,
>>> 8021] for [184.72.197.142/32]
>>> Unable to start the cluster. Terminating all nodes.
>>> java.lang.NullPointerException
>>> at
>>> org.apache.whirr.service.hadoop.HadoopConfigurationBuilder.buildMapReduceConfiguration(HadoopConfigurationBuilder.java:113)
>>>  at
>>> org.apache.whirr.service.hadoop.HadoopConfigurationBuilder.buildMapReduce(HadoopConfigurationBuilder.java:71)
>>> at
>>> org.apache.whirr.service.hadoop.HadoopClusterActionHandler.createHadoopConfigFiles(HadoopClusterActionHandler.java:102)
>>>  at
>>> org.apache.whirr.service.hadoop.HadoopClusterActionHandler.beforeConfigure(HadoopClusterActionHandler.java:88)
>>> at
>>> org.apache.whirr.service.ClusterActionHandlerSupport.beforeAction(ClusterActionHandlerSupport.java:53)
>>>  at
>>> org.apache.whirr.actions.ScriptBasedClusterAction.execute(ScriptBasedClusterAction.java:100)
>>> at
>>> org.apache.whirr.ClusterController.configureServices(ClusterController.java:148)
>>>  at
>>> org.apache.whirr.ClusterController.launchCluster(ClusterController.java:110)
>>> at
>>> org.apache.whirr.cli.command.LaunchClusterCommand.run(LaunchClusterCommand.java:63)
>>>  at org.apache.whirr.cli.Main.run(Main.java:64)
>>> at org.apache.whirr.cli.Main.main(Main.java:97)
>>>
>>> I am now running the integration tests on my machine to make sure that
>>> everything still works fine.
>>>
>>> -- Andrei Savu / andreisavu.ro
>>>
>>> On Wed, Jan 4, 2012 at 12:23 PM, Andrei Savu <[email protected]>wrote:
>>>
>>>> I will give it a try now and get back to you with more feedback.
>>>>
>>>>
>>>> On Wed, Jan 4, 2012 at 2:22 AM, David Dundas <[email protected]>wrote:
>>>>
>>>>> Also just added whir as the cluster-user, and it looks like I'm still
>>>>> getting the same errors.
>>>>>
>>>>>
>>>>> On Tue, Jan 3, 2012 at 7:21 PM, David Dundas <[email protected]>wrote:
>>>>>
>>>>>> Oh I understand now. I definitely misread your post. Ok then what is
>>>>>> the difference between
>>>>>> whirr.cluster-user and whirr.login-user?
>>>>>>
>>>>>> Do I need to create ssh key for the cluster user as well?
>>>>>>
>>>>>> Should I run the command as the cluster user or the command user?
>>>>>>
>>>>>> Can the login-user and cluster-user be the same?
>>>>>>
>>>>>> Thanks!
>>>>>>
>>>>>> On Tue, Jan 3, 2012 at 7:04 PM, Andrei Savu <[email protected]>wrote:
>>>>>>
>>>>>>>
>>>>>>> On Wed, Jan 4, 2012 at 1:57 AM, David Dundas <[email protected]>wrote:
>>>>>>>
>>>>>>>> * have you added something like whirr.cluster-user=whirr to
>>>>>>>> hadoop.properties?
>>>>>>>> >>> No I havent - should I be?
>>>>>>>>
>>>>>>>
>>>>>>> Yes!! Whirr is trying to create the ec2-user on the remote machine
>>>>>>> and it fails because that user already exists.
>>>>>>>
>>>>>>> I have created the following two related JIRA issues:
>>>>>>> * https://issues.apache.org/jira/browse/WHIRR-463
>>>>>>> * https://issues.apache.org/jira/browse/WHIRR-464
>>>>>>>
>>>>>>> As I said already in a comment my advice is to add:
>>>>>>>
>>>>>>> whirr.cluster-user=whirr (to hadoop.properties)
>>>>>>>
>>>>>>> and you should be able to login using something like:
>>>>>>>
>>>>>>> ssh -i ~/.ssh/id_rsa_whirr whirr@remote-host-ip
>>>>>>>
>>>>>>> -- Andrei
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>
>>>
>>
>>
>

Reply via email to