Hi Ronnie
No idea what is going on here, but just being pragmatic:

the bin/janusgraph.sh script starts Cassandra, Elasticsearch and Gremlin Server using the conf/janusgraph-cql-es.conf graph configuration. As you can check in the janusgraph distribution files, this config does not specify the graph.timestamps  property (so defaults to MICRO).

So some things to try:

• does your Cassandra cluster have custom settings for the timestamp type? Then you hit an existing janusgraph issue.
• does your system have a locale different than english-international and does changing the locale recover the issue? Then you discovered a janusgraph issue!
• does removing the graph.timestamps=MICRO line help? Note that you can only try this on a new graph, because this setting is FIXED.

Best wishes,    Marc