[
https://issues.apache.org/jira/browse/WHIRR-117?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tom White updated WHIRR-117:
----------------------------
Attachment: WHIRR-117.patch
Thanks for the feedback, Adrian - it's very helpful! I've updated the patch
with your suggestions, and a few more changes:
* I've introduced a ClusterActionEvent object, to make the ClusterActionHandler
be in line with the standard listener pattern, like ServletContextListener.
* Good point about runurl - I've generalized to Statement (from jclouds). I
hadn't seen all the work you've done on scriptbuilder in jclouds - it looks
very useful, and something we could use more in Whirr (e.g. the ability to
build files via Statements.createFile), so the ability to use any Statement is
a good improvement.
* The Hadoop role names are now backwards compatible (nn, jt, dn, tt), but not
yet fully composable since there is more work to be done it splitting the
cloud-side scripts into individual roles. This can be tackled in a future JIRA.
* ClusterAction implementations are now in a new package
o.a.whirr.cluster.actions.
* DestroyClusterAction is new; now all the common actions are implemented as
ClusterActions.
* ScriptBasedClusterAction is new, and is an implementation of ClusterAction
for actions that run scripts on nodes. In the previous patch this was built
into ClusterAction, which was too tightly coupled - e.g. a PalletClusterAction
would use Pallet directly and not the jclouds script running machinery.
* One area that could be improved is how the cluster objects are updated in
ScriptBasedClusterAction. Actions and handlers may update the cluster object,
and the logic for propagating the change to other events is a little delicate
(although encapsulated in ScriptBasedClusterAction). Since this is essentially
an internal API it could be improved later.
I've done some testing and so far everything passes (on EC2). I'm going to do
some more tests on Rackspace.
> Composable services
> -------------------
>
> Key: WHIRR-117
> URL: https://issues.apache.org/jira/browse/WHIRR-117
> Project: Whirr
> Issue Type: New Feature
> Components: core
> Reporter: Tom White
> Assignee: Tom White
> Fix For: 0.3.0
>
> Attachments: WHIRR-117.patch, WHIRR-117.patch, WHIRR-117.patch
>
>
> The current design does not support composable services, so you can't, for
> example, run ZooKeeper on the same cluster as Hadoop (a better example would
> be running Flume agents on a Hadoop cluster). We should make it easy to do
> this.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
