Aurelius Titan and Elasticsearch CONTAINS_REGEX not giving results - regex

Question

I have a basic Titan instance running on Cassandra with elastic search for full-text indexing (se below for setup).
Elastic search index does not appear to return correctly when using CONTAINS_REGEX when the string parameter does not contain any sepertors like whitespace or ..
Example:
g.V().has('name', "somename") //This exist
==> v[123456]
g.V().has('name', CONTAINS_REGEX, ".*somename.*") //This does not find the vertex.
==> null
g.V().has('name', CONTAINS_REGEX, ".*omenam.*") //Nor does this..
==> null
g.V().has('name', CONTAINS_REGEX, "somename") //Or this..
==> null
g.V().has('name', CONTAINS_REGEX, "^somename$") //String start + end does not work.
==> null
g.V().has('name', "somename.two") //If the name has a separator in it, then CONTAINS_REGEX works as expected
==> v[23456]
g.V().has('name', CONTAINS_REGEX, ".*somename.*")
==> v[23456]
g.V().has('name', CONTAINS_REGEX, ".*omenam.*")
==> v[23456]
g.V().has('name', CONTAINS_REGEX, "somename\\.two")
==> v[23456]
//Changing to the new syntax also does not work as expected.
g.query().has('name', CONTAINS_REGEX, ".*somename.*").vertices()
So, as you can see, CONTAINS_REGEX does not appear to behave correctly when the string does not contain any whitespace or other separators. This has worked on previous versions (0.4.X). Maybe something has changed that I am not aware of, and this should be done differently now. I was not able to find anything in the docs however. It could also be that there are some settings in either titan or ES I have not set up correctly, but again, I could not find anything.
Any help would be greatly appreciated :)
I am inserting the names like this:
def get_or_create_name(g, name)
{
nv = g.V('name', name);
if(!nv.hasNext())
{
ret = g.addVertexWithLabel('name');
ElementHelper.setProperties(ret, 'name', name);
}
else
{
ret = nv.next();
}
ret
}
Using:
Titan 0.5.2
Debian 7.6
Java java version "1.7.0_65", OpenJDK Runtime Environment (IcedTea 2.5.1) (7u65-2.5.1-5~deb7u1), OpenJDK 64-Bit Server VM (build 24.65-b04, mixed mode)
Indexing part of db schema:
name_label = mgmt.makeVertexLabel("name").make()
name = mgmt.makePropertyKey('name').dataType(String.class).make()
name_uniqueness = mgmt.buildIndex("name_uniqueness",Vertex.class).addKey(name).unique().buildCompositeIndex();
mgmt.setConsistency(name_uniqueness, ConsistencyModifier.DEFAULT); //Not sure if this is needed
mgmt.buildIndex('name_index',Vertex.class).addKey(name,Mapping.TEXT.getParameter()).buildMixedIndex("search")
Small note: I am aware that I could use Mapping.STRING.getParameter() in stead of TEXT, and then use g.V().has('name', REGEX, ".*somename.*"), but I was never able to get the string index to work at all. I am also aware CONTAINS_REGEX is intended for use on each word within a longer text, but again, I was never able to get the string version to work.

Answer 1

After digging through error logs and testing lots of stuff I found that the issue was that some of the java-processes were running out of memory. Specifically Rexster.
What I did was increase the available memory and tweak the java options a bit.
Thus, in bin/rexster.sh I changed this line:
JAVA_OPTIONS="-server -Xms128m -Xmx512m -Dtitan.logdir=$LOG_DIR"
To:
JAVA_OPTIONS="-server -Xms128m -Xmx2g -XX:-UseConcMarkSweepGC -Dtitan.logdir=$LOG_DIR"
And now everything seems to be working nicely! :)
To recap: The issue was that Rexster crashed between storing to cassandra and the initialization of the elasticsearch index because of lack of ram.