[
https://issues.apache.org/jira/browse/METRON-552?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15667358#comment-15667358
]
Justin Leet commented on METRON-552:
------------------------------------
As a addendum to this, before the topology can be started, it's also necessary
to create the pcap kafka topic (doesn't need to be populated with data or
anything, but it needs to actually exist to be read from in the topology).
> Ambari Mpack should be able to manage pcap topology
> ---------------------------------------------------
>
> Key: METRON-552
> URL: https://issues.apache.org/jira/browse/METRON-552
> Project: Metron
> Issue Type: Improvement
> Affects Versions: 0.2.1BETA
> Reporter: Justin Leet
>
> Right now, the mpack installs the pcap RPM, but doesn't do any management of
> it. This should be handled by Ambari. With the current boundaries of the
> mpack (i.e. responsibility begins at kafka, not before), using pycapa to
> produce data to Kafka wouldn't be managed by Ambari.
> As a workaround (which will also have to happen in the Ambari service), pcap
> topology can be run on an Ambari managed Metron instance by
> 1) Updating /usr/metron/0.2.1BETA/config/pcap.properties by editing kafka.zk
> manually. Ambari should manage these configs itself, once the service is up.
> 2) Creating /apps/metron/pcap on HDFS, owning it to metron:hadoop, and
> updating permissions to 775 on the folder.
> At this point, /usr/metron/0.2.1BETA/bin/start_pcap_topology.sh should be
> able to be run normally.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
