[
https://issues.apache.org/jira/browse/METRON-573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15673889#comment-15673889
]
Nick Allen commented on METRON-573:
-----------------------------------
If the deployment process were to use the internal DNS names, which do not
change, then this *might* fix the problem. The only consideration being that
these internal names can only be resolved within the same VPC (virtual private
cloud; Amazon-speak).
* Example of an external DNS name that can and will change on reboot;
ec2-35-163-234-95.us-west-2.compute.amazonaws.com
* Example of an internal DNS name that will not change on reboot;
ip-10-0-0-143.us-west-2.compute.internal
> Cluster deployed via 'amazon-ec2' process does not survive reboot
> -----------------------------------------------------------------
>
> Key: METRON-573
> URL: https://issues.apache.org/jira/browse/METRON-573
> Project: Metron
> Issue Type: Bug
> Reporter: Nick Allen
>
> The current AWS cloud deployment process results in a cluster that does not
> survive a reboot. The cluster is configured using the external DNS names.
> These names can change when a host is rebooted. When the DNS names change
> the configuration for all the core components like Storm, HBase, HDFS, etc
> are invalid. None of the core components can even be started from Ambari.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
