[
https://issues.apache.org/jira/browse/TEZ-3488?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15646781#comment-15646781
]
George Liaw commented on TEZ-3488:
----------------------------------
[~Sreenath] The use case would be for those of us operationalizing the
configuration and using templating to do so. For example, jinja templating like:
{noformat}{{ host }}:{{ port }}{noformat}
But as we spoke about earlier, either handling it in the Tez code or explicitly
stating this in the docs would be okay, as long as we know to account for it
when writing our configuration playbooks/recipes. If the port rewrite isn't
handled in the Tez code, we would then rewrite the jinja template as:
{noformat}{{ host }}{{ ':' + port if port == '80' or port == '443' else ''
}}{noformat}
It would be nice if people don't have to consciously worry about the ports when
configuring the cluster if the code or the ops team has already accounted for
it.
> Tez UI Vertex 404
> -----------------
>
> Key: TEZ-3488
> URL: https://issues.apache.org/jira/browse/TEZ-3488
> Project: Apache Tez
> Issue Type: Bug
> Affects Versions: 0.8.4
> Reporter: George Liaw
>
> When we're trying to view vertices for a running job, the Vertex page will
> show a 404 error saying:
> "Could not retrieve expected data from @ <url>"
> And the Vertex Counter page will show:
> "Counters unavailable"
> and
> "Error code: Unknown, message: Resource Manager (RM) out of reach. Either
> it's down or CORS is not enabled"
> The RMs are up and CORS is enabled. If we go to the displayed url, we get the
> expected XML with the application info. When the DAG finishes running, those
> errors go away and the pages display fine. Any idea what's going on here?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
