Neo4j integration: reflection jobs keep failing

#1

Sorry but seems that I am running into issues at every step of this neo4j integration. So in summary here is where I am now…

  1. I connected the JDBC data source per blog and when testing it, I get a success message.
  2. Finally was able to complete the neoNode app script to create all these reflection jobs in Siren. This graph is pretty huge mind you (maybe the problem, I can’t say for sure)
  3. Now… I am trying to run these reflection jobs and everyone I have tried has failed. Even when I try to test one in the UI, the test button is either greyed out or I get datasource is null error (???).

Really not sure where to go with this now. Please advise.

#2

Hello ssealock,
Have you setup the datasource in Management -> Datasources ?
And when you click ‘Test connection’ on the said datasource, does the connection work fine?

image
image.png1282×781 29.4 KB

What are the exceptions you see in failed Reflection jobs?
Regards,
Varun

#3

Yes, of course. I set up the data source and when testing it i get the popup stating it was successful.

Attached are screenshots where you can see the error in the tooltip of the latest run. Also you can see that I cant even test the query. Let me know what else I can provide to narrow this issue down further.

PS When I start ES, I use nohup and saw more errors in there if that might be helpful let me know

07%20AM
Screen Shot 2019-07-29 at 9.36.07 AM.png2280×1346 330 KB

35%20AM
Screen Shot 2019-07-29 at 10.46.35 AM.png2152×1348 481 KB

09%20AM
Screen Shot 2019-07-29 at 10.11.09 AM.png2238×1358 197 KB

#4

It might be that the node is struggling with too many simultaneous Reflection jobs (Which shouldn’t happen). Could you run 1 or two jobs at a time and see how it goes? (You press the green play icon to manually run a job).

#5

This morning, after rebooting the server and restarting everything, I only kicked off 1 job just to see it try to run through its entirety and it still failed. Images of the nohup log attached…

37%20AM
Screen Shot 2019-07-29 at 10.55.37 AM.png2468×624 180 KB

47%20AM
Screen Shot 2019-07-29 at 10.55.47 AM.png2458×660 219 KB

#6

Could you increase the thread_pool settings in elasticsearch.yml to something like:

thread_pool.write.size: 9
thread_pool.write.queue_size: 1000

Tinker with them and see how it works for you.

#7

Hello ssealock,
Did you managed to get it working eventually? This was an issue with 10.2.x but in 10.3.x (yet to be released, as of today) we have moved to our separate thread_pool rather than using Elasticsearch one for reflection jobs so such events and custom configuration would not be required in future.
Apologies for the inconvenience, let me know if you need anymore help. We can possibly setup a call if required.

Regards,
Varun

#8

HI Varun,
I made the adjustments per your recommendations and was able to advance on a few out of many jobs that were created. Ultimately I stopped evaluating it altogether as I didn’t have much more time to invest. Ill try to resume finishing all the jobs sometime over the weekend and let you know how it goes. Thanks for following up with me.

Regards,
Sue

1 Like
#9

@ssealock 10.3.0 is out and you may refer to the Neo4j Import wizard documentation over here: https://docs.support.siren.io/10.3.0/platform/en/siren-investigate/data-reflection/importing-neo4j-data.html