Setup JanusGraph DB - Githubissues

briri commented 5 years ago

Install JanusGraph onto dmp-dev.

[x] Install Java 8
[x] Install JanusServer
[x] Install Berkeley DB which will support < 100 million vertices/indexes so should be fine for our prototype - https://docs.janusgraph.org/latest/bdb.html
[x] Install ElasticSearch
[x] Configure Janus Server to work with BDB and ES
[x] Install Gremlin console (server-side?)

jimvanderveen commented 5 years ago

Hmm, I thought those were checkboxes above. Anyway, Java 8 is installed, and db4-devel (Berkeley DB) was already installed. There is no 'janus' package in any of our configured repositories.

briri commented 5 years ago

thanks Jim!

briri commented 5 years ago

Installed maven binary (a requirement to build janusgraph) and built janusgraph from source in the /apps/dmp/local directory. According to the JanusGraph docs, it comes with a built in ElasticSearch engine so we do not need to install for this prototype. If we discover that we need a more dedicated indexing engine we can install ElasticSearch locally.

briri commented 5 years ago

Janusgraph is installed and responding on dmp-dev. I also added the cypher-for-gremlin-server plugin so that the gremlin-server running on the service will accept cypher queries.

Decided to just use the built-in Cassandra and ElasticSearch instead of BerkeleyDB

Everything is installed under the local dmp role account in ~/local/janusgraph-0.3.0-hadoop2
Custom config files are ./conf/http-gremlin-server.yaml and ./conf/http-janusgraph-cql-es-server.properties
Custom startup script is ./bin/http-janusgraph.sh. This file starts up Cassandra, ElasticSearch and GremlinServer usong the standard start/stop/status commands. (This should be kicked off on server startup at some point)
Created a custom apache vhost for this that was manually swapped in to replace the existing vhost. It will need to swapped back in when instance restarts. janusgraph-dmptool-dev.cdlib.org.conf

CDLUC3 / maDMPs

Setup JanusGraph DB #9