[
https://issues.apache.org/jira/browse/AMBARI-19429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15929306#comment-15929306
]
Roman Shaposhnik commented on AMBARI-19429:
-------------------------------------------
[~jluniya] thanks for providing the patch! It seems to be working for the
service that rely on common services definitions (HDFS and ZOOKEEPER) but fails
for the custom services (like YARN and HIVE). The failure seems to have to do
with agents not being synchronized with the stack definition. E.g. here's how
the YARN fails:
{noformat}
Caught an exception while executing custom service command: <class
'ambari_agent.AgentException.AgentException'>: 'Script
/var/lib/ambari-agent/cache/stacks/ODPi/2.0/services/YARN/package/scripts/nodemanager.py
does not exist'; 'Script
/var/lib/ambari-agent/cache/stacks/ODPi/2.0/services/YARN/package/scripts/nodemanager.py
does not exist'
Command failed after 1 tries
{noformat}
and indeed that script is missing. In fact, on agent side all of services are
missing. E.g.
{noformat}
$ ls -l /var/lib/ambari-agent/cache/stacks/ODPi/2.0
total 4
drwxr-xr-x. 7 root root 4096 Mar 17 01:02 hooks
{noformat}
As you can see there are no services subdirectory there -- just hooks.
Can you help figure out what's going on here?
> Create an ODPi stack definition
> -------------------------------
>
> Key: AMBARI-19429
> URL: https://issues.apache.org/jira/browse/AMBARI-19429
> Project: Ambari
> Issue Type: Improvement
> Components: stacks
> Affects Versions: 2.4.2
> Reporter: Roman Shaposhnik
> Assignee: Roman Shaposhnik
> Fix For: trunk
>
> Attachments: AMBARI-19429-mpack_trunk.patch, AMBARI-19429.patch2.gz,
> AMBARI-19429.patch.gz, AMBARI-19429_trunk.patch
>
>
> ODPi is a nonprofit organization committed to simplification &
> standardization of the big data ecosystem with common reference
> specifications and test suites. As part of its mission, ODPi has been
> developing a series of specifications for how to integrate upstream Apache
> projects into the coherent platform. Part of this standardization effort is
> maintenance of the ODPi core stack definition which today includes:
> * Apache Zookeeper
> * Apache Hadoop
> * Apache Hive
> and has been maintained as a custom stack on ODPi side:
>
> https://github.com/odpi/bigtop/tree/odpi-master/bigtop-packages/src/common/ambari/ODPi/1.0
> In conjunction with merge effort for Apache Bigtop BIGTOP-2666 I'd like to
> propose that instead of migrating the stack definition to Bigtop, we should
> actually migrate it to Ambari.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
