I was trying to get Cassandra 0.6.8 (latest stable release) going for
the first time and my attempts at getting the example code to run caused
Cassandra to die with out a JavaMemoryError just by connecting
cassandra-cli to cassandra.
java.lang.OutOfMemoryError: Java heap space
at
org.apache.thrift.protocol.TBinaryProtocol.readStringBody(TBinaryProtocol.java:296)
at
org.apache.thrift.protocol.TBinaryProtocol.readMessageBegin(TBinaryProtocol.java:203)
at
org.apache.cassandra.thrift.Cassandra$Processor.process(Cassandra.java:1154)
at
org.apache.cassandra.thrift.CustomTThreadPoolServer$WorkerProcess.run(CustomTThreadPoolServer.java:167)
at
java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:662)
Looking at the thrift code it is allocating arrays based on lengths read
of the wire, without adequate validation of the length. This allows
client errors to crash the server :(
The steps were:
1) I was trying to use the Java sample client code in
http://wiki.apache.org/cassandra/ThriftExamples
I got past the pains of thrift install, and the fact that the sample
code is out of date. Constructor signatures have changed in the code
thrift r917130 (as used in 0.6.8) generates.
2) I then ran into the problem that the sample client uses framed
transport, but cassandra default storage-conf.xml uses unframed
transport. This didn't give a clean error, but just cause the client to
block for ever :(
I changed the cassandra storage-conf.xml to use framed transport to
resolve this.
3) Using the cassandra-cli to connect to cassandra produced the error
above. This is presumably because the client was using unframed
transport and the server wasn't. That said the server should have enough
validation to avoid crashin because of bad clients.
I plan to persevere, but I thought I'd let you all know as these sort of
problems could put off a lot of newcomers.
Nick Reeves
