Cannot install hdp-multinode-default

[sequenceiq]

Cannot install hdp-multinode-default

[keyki]

Unfortunatelly Ambari often fails to install Ganglia due to network issues, I suggest you to try again or remove that service from the blueprint.

[leebrooks0]

I have tried quite a few times, so I am going to strip it from the
blueprint and see what happens

On Thu, Oct 9, 2014 at 11:39 AM, Krisztian Horvath <[email protected]

wrote:

Unfortunatelly Ambari often fails to install Ganglia due to network
issues, I suggest you to try again or remove that service from the
blueprint.

—
Reply to this email directly or view it on GitHub
#15 (comment)
.

[keyki]

Use the amb-shell function and you'll be able to issue the commands manually similar to this except using a custom blueprint with blueprint add --url <url_to_the_blueprint> or blueprint add --file <file_location>

[leebrooks0]

Thanks, that is what I have been doing. A quick question, what is the best way to create 6 dockers with no blueprint?

I have been doing this: amb-deploy-cluster 6 bogus so that it creates the dockers and then fails on cluster creation, whereupon I switch to ambari-shell for the rest

[keyki]

amb-start-first
amb-start-node 1
amb-start-node 2
..

[leebrooks0]

Thanks, another thing has cropped up

I am trying to run this blueprint (stock minus Ganglia and Nagios): https://gist.githubusercontent.com/leebrooks0/3b107c58e85506e6499d/raw/0fe2cb1cf3aa783d5322d60b5ba4b0b87f756cd4/lee-blueprint

CLUSTER_BUILD:no-monitoring>cluster autoAssign
Command failed java.lang.reflect.UndeclaredThrowableException

Here is what my shell looks like:

ambari-shell>blueprint add --url https://gist.githubusercontent.com/leebrooks0/3b107c58e85506e6499d/raw/0fe2cb1cf3aa783d5322d60b5ba4b0b87f756cd4/lee-blueprint
Blueprint: 'no-monitoring' has been added
ambari-shell>cluster build --blueprint no-monitoring
  HOSTNAME         STATE
  ---------------  -------
  amb1.mycorp.kom  HEALTHY
  amb0.mycorp.kom  HEALTHY
  amb4.mycorp.kom  HEALTHY
  amb3.mycorp.kom  HEALTHY
  amb2.mycorp.kom  HEALTHY
  amb5.mycorp.kom  HEALTHY

  HOSTGROUP  COMPONENT
  ---------  ------------------
  master_3   ZOOKEEPER_SERVER
  master_3   RESOURCEMANAGER
  master_2   YARN_CLIENT
  master_2   HIVE_CLIENT
  master_2   HDFS_CLIENT
  master_2   HIVE_SERVER
  master_2   HIVE_METASTORE
  master_2   HISTORYSERVER
  master_2   ZOOKEEPER_CLIENT
  master_2   WEBHCAT_SERVER
  master_2   MYSQL_SERVER
  master_2   SECONDARY_NAMENODE
  slave      NODEMANAGER
  slave      HBASE_REGIONSERVER
  slave      DATANODE
  master_1   YARN_CLIENT
  master_1   HDFS_CLIENT
  master_1   NAMENODE
  master_1   HBASE_MASTER
  master_1   ZOOKEEPER_SERVER
  master_1   HCAT
  gateway    YARN_CLIENT
  gateway    HIVE_CLIENT
  gateway    HDFS_CLIENT
  gateway    SQOOP
  gateway    HBASE_CLIENT
  gateway    OOZIE_CLIENT
  gateway    AMBARI_SERVER
  gateway    PIG
  gateway    ZOOKEEPER_CLIENT
  gateway    MAPREDUCE2_CLIENT
  gateway    HCAT
  master_4   OOZIE_SERVER
  master_4   ZOOKEEPER_SERVER

CLUSTER_BUILD:no-monitoring>cluster autoAssign
Command failed java.lang.reflect.UndeclaredThrowableException

[keyki]

The reason of the exception is not properly shown I'll fix that. The problem is if you want to use the auto assignment at least 1 host group's name should start with slave_ e.g. slave_1.

[leebrooks0]

Is the auto assignment part of the Ambari api?

[keyki]

No, it isn't, but they plan to add such functionality.

[keyki]

We implemented the auto assignment mostly for our needs, but you can always assign manually.

[leebrooks0]

I am now trying this blueprint https://gist.githubusercontent.com/leebrooks0/3b107c58e85506e6499d/raw/db437778c6ddec3bf86f0733d5127ee7cc472728/lee-blueprint which is stock less Nagios, Ganglia , HBase and Oozie.

I get this error with Hive:

This is the error I get Python script has been killed due to timeout

Originally the above blueprint had included HBase and Oozie, but I removed them as they were getting this error too.

Am I doing something wrong or is the blueprint API not ready for use yet?