[
https://issues.apache.org/jira/browse/HTRACE-144?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14381029#comment-14381029
]
Colin Patrick McCabe commented on HTRACE-144:
---------------------------------------------
Another quick note about choosing the best ip address:
To handle multi-homing:
* do not use localhost addresses (i.e. 127.0.0.x) if at all possible.
* prefer non-routable addresses (i.e. internal ones, such as 192.168.0.1,
etc.). These should be used by nodes to talk to each other, rather than the
internet at large.
* if all else fails, pick the non-localhost IP address which comes first in
sorting order, to ensure that we at least have a consistent identifier.
* if everything is totally screwed (can't access interfaces, etc.), return
127.0.0.1 (shouldn't happen)
I think in 99% of cases, this will be adequate. even in multi-homing cases.
If this isn't adequate, the user can configure a specific process ID per node
by setting the {{process.id}} configuration variable to a non-default value.
I also think we should add a {{$\{pid\}}} variable later, to inject the UNIX
process ID. However, this turns out to be complicated in Java so let's do it
in a follow-on. Same for hostnames, I think... we should be careful there
since there are hairy DNS issues we'll need to think about with hostnames
> Include IP address in span process description
> ----------------------------------------------
>
> Key: HTRACE-144
> URL: https://issues.apache.org/jira/browse/HTRACE-144
> Project: HTrace
> Issue Type: Improvement
> Affects Versions: 3.2.0
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Attachments: HTRACE-144.001.patch, HTRACE-144.002.patch
>
>
> Right now, we have no IP address or hostname specified in the span process
> description. So if we have 100 datanodes, the spans from all of them simply
> show up as "DataNode." Obviously this is unsatisfactory.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
