not a support question
txie
Member
I set up 4 nodes cassandra cluster in one DC, which worked fine yesterday. Today I decided to set it up as a multiple-DC cluster, and made a bunch of changes in cassandra.yaml. Now things get worse.
1. I run 'service dse start' to start cassandra, however after a few seconds, the cassandra java process exits with no error stack.
2. I don't know if my old configuration is still cached or kept in certain ways, what's the best way to clean things up, and give it a fresh start?
Thanks,
Tao
Instead of using "service dse start" can you use "<installdir>/bin/dse cassandra". If you are running a pure Cassandra system (no DSE, sorry I forgot there are people doing that ;-) use "<installdir>/bin/cassandra -f" to run it.
1) Is there anything shown on the console?
2) There is no output in /var/log/cassandra/system.log et al?
3) Do you see anything in the system log if you set the log4j "log4j.rootLogger" to be at INFO level?
4) Is there any number printed when you run 'pgrep -f cassandra'? If yes, you may have an already running server process.
Thanks,
Sven
txie
Member
Hi Sven,
/usr/bin/dse cassandra makes the cassandra server started without exit. system.log shows the error:
java.lang.RuntimeException: Unable to create thrift socket to /15.185.165.124:9160
at com.datastax.bdp.server.DseDaemon$ThriftServer.<init>(DseDaemon.java:271)
at com.datastax.bdp.server.DseDaemon.startServer(DseDaemon.java:168)
Looks like I shouldn't have set the rpc_address to the public IP, after changing to the private ip of the node, things started working. nodetool and cassandra-cli both can work on the node.
Thanks a lot for your help!
Tao
Great, glad to hear that.
Sven
