0

I am setting up a 3 node HA Neo4j cluster on Eucalyptus (which is similar to using AWS). All the nodes come up just fine (in the default standalone mode which you can see in the logs), but after restarting the neo4j service in HA mode, I get the following error in the /var/log/neo4j/neo4j.0.0.log. Below contains the error, and configuration of each node thats part of the HA Neo4j cluster:

https://gist.github.com/4657797

Any ideas as to why this error is occurring? Thanks ahead of time for any help in trying to resolve this issue.

Cheers,

Harold Spencer, Jr.

hasp77
  • 61
  • 2
  • This looks like a classpath or logging config problem. Have you tried 1.9.M04 instead which has less moving parts with the Zookeeper-less config? – Peter Neubauer Jan 28 '13 at 20:05
  • I think you're currently mixing the new HA setup and the old one, the new one doesn't need zookeeper, it seems you're running into a classpath issue it can find the libraries for the log4j logging framework, can you check the contents of system/coordinator/lib ? And can you please also share the content of the `data/graph.db/messages.log` files? – Michael Hunger Jan 29 '13 at 09:03
  • there was a change in packaging sorry for that, can you please copy system/coordinator/lib/* to system/lib and try again? Thx – Michael Hunger Jan 29 '13 at 09:43
  • Hey guys, Thanks for the feedback. I will do what Michael suggested and get back to you guys. Thanks again for this. Cheers - – hasp77 Jan 29 '13 at 12:21

1 Answers1

0

Harold, it seems that you still have configured Zookeeper, which fails to start. If you follow http://docs.neo4j.org/chunked/snapshot/ha-setup-tutorial.html, you should get a Zookeeper-free setup, resulting in something like this https://gist.github.com/4663136

Feel free to follow up with more details, I even made a screencast on this, see https://vimeo.com/51906007

HTH

/peter

Peter Neubauer
  • 6,311
  • 1
  • 21
  • 24
  • Hi Peter, thanks for the info. It seems I was doing a bit overkill. I am testing it out now. I will keep you posted to my findings. Thanks again for your help with this. Cheers - – hasp77 Jan 29 '13 at 17:00
  • Hey Peter, I got it working! Thanks so much. I just have to do some cleanup. – hasp77 Jan 29 '13 at 21:25
  • Hey Peter, I got it working! Thanks so much. I just have to do some cleanup. I have one question though, I ran curl to query the cluster and everything looks good. https://gist.github.com/4668102 The only thing is that when I go to the UI, it only says that I have one node under the Dashboard - even though under the Server Info tab, and the JSON output from the curl query show all three. I thought it had to do with the backup uris not showing up for all three nodes. My configs are below: https://gist.github.com/4668281 Look forward to hearing from you soon. Cheers -- – hasp77 Jan 29 '13 at 22:13
  • Hi Peter, Looking at my setup, I have all the cluster communication traffic being handled on private IPs, and the web service ports tied to the public IPs. Could this be a reason why the Dashboard UI shows only 1 node? Cheers - – hasp77 Jan 30 '13 at 12:41