It appears to read the CSV file just fine, but seems to be choking on the "hadoop_namenode_host" variable --- probably this is a puppet variable scoping issue that I don't understand very well, or possibly a bug (but I doubt it since puppet recipes work just fine in vagrant).
Also, maybe it could be a puppet versioning thing? Im using puppet 3.6.2
Notice: Scope(Class[Hadoop::Common-hdfs]): in common-hdfs <<<<< Could not find value for 'hadoop_namenode_host'
So that is what breaks the parser : I guess it wants that value ("hadoop_namenode_host") to be defined, in order to properly parse the template.
But it IS defined in the datanode snippet, just not in the common-hdfs puppet snippet.
So some how that variable "loses" its scoping or definition in init.pp but im not sure how. I am suspecting maybe the version of puppet is important here, or else, maybe the ordering of components, or some other system specific aspect of my config (I DONT think this could be a bug, given that so many people are succesfully using bigtop puppet recipes).
On Thu, Jul 3, 2014 at 11:40 PM, jay vyas <[EMAIL PROTECTED]> wrote: jay vyas
The version is important. 3.x is VERY different in many language respects. What I do - at least until 3.x isn't supported by our recipes - is just add epel repo to my centos VMs and install Puppet 2.7.8 or whatever is there. Well, at least there's EPEL ;(
HEY BIGTOP ! I figured it out ! SEE BIGTOP-1365 for details (puppet 3 isnt supported yet !) duh :). It turns out the way we reference variables using 2.7 syntax is INCOMPATIBLE with the way puppet 3 references variables (reuires fully qualified class path). On Fri, Jul 4, 2014 at 10:14 AM, jay vyas <[EMAIL PROTECTED]> wrote: jay vyas
NEW: Monitor These Apps!
All projects made searchable here are trademarks of the Apache Software Foundation.
Service operated by Sematext