0

I am attempting to get a DSE (graph and Spark) cluster up and running for a technology pilot.

I am attempting to run a cluster with 2 data centres and two nodes in each data centre.

I used the following instructions to install DSE 5.0.1 into a 4 separate Debian VMs running on my Mac. [http://docs.datastax.com/en/datastax_enterprise/5.0/datastax_enterprise/install/installDEBdse.html]

I looked through the instructions and the dse.yaml configuration file and didn't see anything I needed to edit that file. so I moved onto the cassandra.yaml configuration file.

In that I altered;

  • cluster_name 'My Test'

  • seeds "192.168.0.201, 192.168.0.202, ...203, ...204"

(i added all 4 nodes to the seeds list - so that I can remove and and add any of them - again for the pilot - I am aware that I would normally have more nodes - and would not have all nodes listed as seed nodes.)

*endpoint_snitch: GosspingPropertyFileSnitch

and then I edited the /etc/dse/cassandra/cassandra-rackdc.properties

so that it read (for two nodes);

dc=My_Test_DC1

rack=My_Test_Rack1

and it read ... Rack2 for the last two nodes.

I have edited all four configs to be identical and when I attempted to start dse on the very first node I get the following error;

ERROR 18:54:28,983 CassandraDaemon.java:698 - Fatal exception during initialization org.apache.cassandra.exceptions.ConfigurationException: Saved cluster name Test Cluster != configured name My Test

If I haven't yet run anything, how can it be complaining that I have the wrong name?

So, I found this;

[cassandra - Saved cluster name Test Cluster != configured name

And changed the cluster name back to Test Cluster in cassandra.yaml and attempted to restart the node again.... but got a further error;

ERROR 19:08:43,817 CassandraDaemon.java:698 - Cannot start node if snitch's data center (My_Test_DC1) differs from previous data center (Cassandra). Please fix the snitch configuration, decommission and rebootstrap this node or use the flag -Dcassandra.ignore_dc=true.

So I changed the DC and the Rack names back to the default - and now it works appropriately.

So I suppose ultimately what I am after is - how do I go about starting DSE with graph and Spark - AND use my configuration?

Is there a "recipe" somewhere I can follow? is there something obvious I have missed?

As always - Thanks! -Gavin.

Community
  • 1
  • 1
Gavin Baumanis
  • 343
  • 1
  • 11

1 Answers1

0

I found two things that I had correct.

First was delete existing data, which came with the install sudo rm -rf /var/lib/cassandra

And I also found that I did not have the LISTEN address or the RPC address set correctly.

Once I did these, all four nodes joined the cluster successfully.

And I am not sure if it was causing any issues or not - but I removed two of the seed addresses and just had one in each rack.

-Gavin.

Gavin Baumanis
  • 343
  • 1
  • 11