Note that Pig is a client-side program, so nothing further needs installing on the cluster. Hive can work that way too, but it would be better to have the Hive metastore installed on the cluster (see https://issues.apache.org/jira/browse/WHIRR-343).
It would also be very useful I think to have a role that installs the client side programs so that you can SSH into that node and run programs from there. This was discussed a little bit in the context of Mahout here: https://issues.apache.org/jira/browse/WHIRR-384?focusedCommentId=13103312&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13103312 Cheers, Tom On Fri, Oct 28, 2011 at 7:24 AM, Stephen Boesch <[email protected]> wrote: > Hi Andrei, > Thanks for the tip, i'm looking at the presentation now. Should be enough > to get started. > > 2011/10/28 Andrei Savu <[email protected]> >> >> Check the presentations page on the wiki and take a look at >> services/zookeeper - it should be self explanatory. >> >> Feel free to ask any question! >> >> On Oct 28, 2011 4:56 PM, "Stephen Boesch" <[email protected]> wrote: >>> >>> Hi Andrei, >>> Thanks for the tip about not running as root / specifying the >>> whirr.user. >>> As for these recipes: i'd be more than glad to put them together - do >>> you have any pointers on how to add new recipes? I started to look through >>> the 0.60-incubating but would be faster if you were to write a sentence or >>> two pointing the way. >>> thx! >>> stephenb >>> >>> 2011/10/28 Andrei Savu <[email protected]> >>>> >>>> Stephen, >>>> >>>> Hive and Pig are not yet available as Whirr services but you should be >>>> able to easily deploy them on a Hadoop cluster started by Whirr. >>>> >>>> I encourage you to contribute back if you add them :) >>>> >>>> Cheers, >>>> Andrei >>>> >>>> On Oct 28, 2011 4:01 PM, "Stephen Boesch" <[email protected]> wrote: >>>>> >>>>> anyone have these? thx! > >
