Github user merrimanr commented on the issue:
https://github.com/apache/metron/pull/934
I agree with @justinleet. I would prefer we use Solr 6+. At some point
HDP Search will move to 6+ and we can easily switch to the Mpack since
installing Solr is a manual step now. Happy to document that once we are all
in agreement.
As for the parameters we expose in Ambari, I'm referring to the parameters
stored in Zookeeper that the SolrWriter reads. They are:
- solr.zookeeper
- solr.commitPerBatch
- solr.commit.waitSearcher
- solr.commit.waitFlush
- solr.commit.soft
- solr.collection
- solr.http.config
Currently Ambari only exposes the "solr.zookeeper" parameter which is
necessary to get the indexing topology writing to Solr. I think a reasonable
solution would be to include all the others except "solr.http.config" since
it's an advanced config. Storm tuning parameters are independent of the writer
implementation and already exposed in Ambari.
For the collection service actions, it would mirror what we expose for ES:
creating and deleting templates (schemas in Solr terms).
---
