Cannot startup JanusGraph when using remote elasticsearch

[ericsodt]

We previously had JG up and running without any issues using HBASE as our backend storage. To try and speed up queries we stood up an elasticsearch cluster, however now our JG instance is not starting up, saying that "could not get type for name org.janusgraph.example.GraphApp".

Our elastic properties are setup like:

index.search.backend="elasticsearch"
index.search.hostname = "myhost.elasticsearch:9200"  ## https://myhost.elasticsearch:9200/janusgraph  <-- returns index values
index.search.elasticsearch.client-only=true

I've been banging my head on this one for a while. Is there anything that I am missing??

[vtslab]

JanusGraph does not handle quotes in the properties files, so please try without.

[ericsodt]

Thanks, I tried that but got the same outcome

[vtslab]

Did you take into account the mutability levels of the graph?
Also beware of lurking failed instances of the graph.

[ericsodt]

With the Gremlin console open, I am creating a new instance with a new hbase table, which previously did not exist. It's during the call to .open() that I am getting these errors. I'm not sure if mutability is an issue in this case as it should not have any instances running.

[vtslab]

Indeed, for a new graph with the HBase table nonexistent, mutability levels are not relevant. So, two more questions:

1. can you provide the entire properties file that you use?
2. can you still open a new graph with just HBase and no configs for elasticsearch?

Also, current JanusGraph versions do not seem to use the index.search.elasticsearch.client-only property anymore and do not have explicit default 9200 port number in example configs, but I do not expect that to be an issue.