Ran,
I ran the nodeprobe tool , and got the correct ring infomation.
Address Status Load Range
Ring
85080781597816482766914734169501403890
192.168.13.40 Up 38.18 GB 26049930823101849079767692175708865961
|<--|
192.168.13.39 Up 38.18 GB 85080781597816482766914734169501403890
|-->|
Thanks!
cheers,
2010-03-17
casablinca126.com
发件人： Ran Tavory
发送时间： 2010-03-17 11:31:41
收件人： user@cassandra.apache.org
抄送：
主题： Re: failover exception with hector
I seems the host you were connected to was unable to locate the other host in
the ring. That's probably a configuration error.
Can you run nodeprobe -host1 ring
where host1 is the host hector was connected to initially?
If not that, can you check with jconsole how hector sees the ring: connect to
the java client running hector and check me.prettyprint.hector if the list of
known hosts is consistent with your setup?
2010/3/16 casablinca126.com <casabli...@126.com>
hi folks,
Just switched from C++ interface to hector, found an exception when
testing with a 2-node cluster:
Exception in thread "main" org.apache.thrift.TException:
Unable to failover to next host
at
me.prettyprint.cassandra.service.KeyspaceImpl.skipToNextHost(KeyspaceImpl.java:598)
at
me.prettyprint.cassandra.service.KeyspaceImpl.operateWithFailover(KeyspaceImpl.java:668)
at
me.prettyprint.cassandra.service.KeyspaceImpl.insert(KeyspaceImpl.java:270)
at
me.prettyprint.cassandra.service.ExampleClient.insertFromDB(ExampleClient.java:108)
at
me.prettyprint.cassandra.service.ExampleClient.main(ExampleClient.java:45)
I was doing a lot of insertion. Probably inserting to 1 node wsa failed,
and hector failed to find the other node.
I'm using cassandra-0.5.0. In conf/storage-conf.xml, the ListenAddress
is configured as the IP and ThriftAddress
is left blank.
Any ideas?
cheers,
Cao Jiguang
--------------
casablinca126.com
2010-03-17
