You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@whirr.apache.org by "Roman Shaposhnik (JIRA)" <ji...@apache.org> on 2013/01/30 19:57:12 UTC
[jira] [Comment Edited] (WHIRR-681) enhance puppet service with an
ability to export cluster topology to the puppet code
[ https://issues.apache.org/jira/browse/WHIRR-681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13566771#comment-13566771 ]
Roman Shaposhnik edited comment on WHIRR-681 at 1/30/13 6:55 PM:
-----------------------------------------------------------------
Re: CreateSitePpAndApplyRolesTest -- absolutely! Attached is a new patch with the unit test updated. Sorry for not updating it to begin with.
As to how this feature is going to be used (and perhaps we need to put this in docs some place) -- it allows individual puppet agents discover the topology of the cluster (IOW, what classes are applied to what nodes) and makes decisions based on that. Here's how it will work on the puppet side (using Zookeeper as an example):
{noformat}
$ensemble = extlookup('zookeeper::server', ['127.0.0.1'])
...
and later on in the configuration file template:
<% ensemble.each_with_index do |server,idx| %>
server.<%= idx %>=<%= server %>
<% end %>
{noformat}
That's the general idea.
Now, I'm still tweaking this patch as far as the best integration with Puppet is concerned (hence not putting it into a 'Patch Submitted' state but simply asking for early feedback). I'm rewriting Bigtop Puppet code as part of this as well.
Finally, as for WHIRR-516 -- this JIRA is required to make progress on WHIRR-516. My ultimate goal as far as WHIRR-516 is concerned is to be able to deploy Bigtop clusters by a simple:
{noformat}
whirr.instance-templates=1 puppet:bigtop-hadoop-hdfs::namenode+\
puppet:bigtop-hbase::master+\
4 puppet:bigtop-zookeeper::server+\
puppet:bigtop-hadoop-hdfs::datanode+\
puppet:bigtop-hbase::server
{noformat}
Hope this answers your questions and, once again, thanks for the feedback -- much appreciated!
P.S. Feel free to add me to the project so I can assign these JIRAs to myself.
was (Author: rvs):
Re: CreateSitePpAndApplyRolesTest -- absolutely! Attached is a new patch with the unit test updated. Sorry for not updating it to begin with.
As to how this feature is going to be used (and perhaps we need to put this in docs some place) -- it allows individual puppet agents discover the topology of the cluster (IOW, what classes are applied to what nodes) and makes decisions based on that. Here's how it will work on the puppet side (using Zookeeper as an example):
{noformat}
$ensemble = extlookup('zookeeper::server', ['127.0.0.1])
...
and later on in the configuration file template:
<% ensemble.each_with_index do |server,idx| %>
server.<%= idx %>=<%= server %>
<% end %>
{noformat}
That's the general idea.
Now, I'm still tweaking this patch as far as the best integration with Puppet is concerned (hence not putting it into a 'Patch Submitted' state but simply asking for early feedback). I'm rewriting Bigtop Puppet code as part of this as well.
Finally, as for WHIRR-516 -- this JIRA is required to make progress on WHIRR-516. My ultimate goal as far as WHIRR-516 is concerned is to be able to deploy Bigtop clusters by a simple:
{noformat}
whirr.instance-templates=1 puppet:bigtop-hadoop-hdfs::namenode+\
puppet:bigtop-hbase::master+\
4 puppet:bigtop-zookeeper::server+\
puppet:bigtop-hadoop-hdfs::datanode+\
puppet:bigtop-hbase::server
{noformat}
Hope this answers your questions and, once again, thanks for the feedback -- much appreciated!
P.S. Feel free to add me to the project so I can assign these JIRAs to myself.
> enhance puppet service with an ability to export cluster topology to the puppet code
> ------------------------------------------------------------------------------------
>
> Key: WHIRR-681
> URL: https://issues.apache.org/jira/browse/WHIRR-681
> Project: Whirr
> Issue Type: Improvement
> Components: service/puppet
> Affects Versions: 0.8.1
> Reporter: Roman Shaposhnik
> Attachments: WHIRR-681-2.patch.txt, WHIRR-681.patch.txt
>
>
> I would like to propose that we enhance puppet service (btw, there seems to be no service/puppet in the components list) by implement the capability that exports the cluster topology via extlookup queryable manifest that gets added to all the nodes same way site.pp does.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira