Hello, 

What configuration steps we need to follow to enable g.E().count()? 

I am running Janusgraph full package (with Cassandra and Elastic search backend). 

With less than 10000 edges, I am getting time out. 

It seems we can alter  -Xmx  in gremlin.sh, anything else that can help?

Thanks,

Yingjie

Hi Yingjie,

Can you try with:

g.V().in().count()

It probably does not make a difference, but this is how JanusGraph should carry out your query given the way edges are stored.

Roughly speaking, the query should take about 1-10 ms per vertex, depending on the performance of your Cassandra cluster and the number of vertices that Cassandra has cached. From your figures I cannot read the number of vertices. But you could try and use the Cassandra nodetool to get some stats on the response times.

HTH,     Marc

Op maandag 14 september 2020 om 14:30:18 UTC+2 schreef Yingjie Li:

Here are the config that I have changed :  

scriptEvaluationTimeout: 300000  ( from default 30000)

-Xms1024m -Xmx8192m (from default: -Xms32m -Xmx512m)

And the problem still holds

 

On Sunday, September 13, 2020 at 11:04:13 AM UTC-4 Yingjie Li wrote:

Hello, 

What configuration steps we need to follow to enable g.E().count()? 

I am running Janusgraph full package (with Cassandra and Elastic search backend). 

With less than 10000 edges, I am getting time out. 

It seems we can alter  -Xmx  in gremlin.sh, anything else that can help?

Thanks,

Yingjie