Hi,

I’ve been working on a storm topology. Our cluster is setup with storm bolts 
running on separate nodes from Hbase nodes. The configs are setup right now, so 
they match the services running on the node. Is this a best practice or should 
/etc/habse/conf/* exist on all nodes regardless?

The problem is that the storm bolt is looking for that file to pickup it’s 
zookeeper configs and others and they aren’t there. Configuration objects 
aren’t serializable, so we’re resorting to cherry picking properties from 
hbase-site.xml and serializing them off to the bolt.

Any advice would be appreciated.

Thanks,

Ian

Reply via email to