Thanks Jonathan,
I found more cases where I wrongly use getInt() directly.
it caused some random data errors: nio.BufferUnderflow. but I did rewind()
these buffers before reading them. then the most probable cause is that some
other threads are sharing the same buffers.
then this prompts me to another question: in current cassandra daemon, is it
possible that a column (or more specifically its bytebuffer) is still being
held by the read
thread , while it was partially written by a write thread? this way the read
thread could possibly get garbled result.
Thanks
Yang
On Fri, Jun 10, 2011 at 7:09 PM, Jonathan Ellis <jbel...@gmail.com> wrote:
> Don't use destructive operations on the bytebuffer, always use e.g.
> getLong(buffer.position)
>
> On Fri, Jun 10, 2011 at 8:47 PM, Yang <teddyyyy...@gmail.com> wrote:
> > I'm using thrift.CassandraServer directly within the same cassandra JVM
> to
> > accomplish my application tasks.
> > (I understand that this is not the normal usage mode...., but the error
> here
> > may also be appearing in Cassandra server code
> > development, so I thought it could be of some value to look at )
> > I ran into some issues when I try to parse out values from column name
> > ByteBuffer:
> >
> > something like the following
> > List<ColumnOrSuperColumn> cols = cassandra_svr.get_slice(key, path,
> > predicate, ConsistencyLevel.ONE);
> > for(ColumnOrSuperColumn colOrSup: cols) {
> > Column col = colOrSup.column;
> > long ts = col.name.getLong(); <======
> >
> > sometimes the last sentence gives error "ByteBuffer underflow", I checked
> > that the buffer size is 8 bytes, and the "pos" points to 8 already. it
> > seems that somewhere in
> > thrift.CassandraServer line 179---195
> > public List<ColumnOrSuperColumn> thriftifyColumns(Collection<IColumn>
> > columns, boolean reverseOrder)
> > {
> > ...........................................................
> > {
> > Column thrift_column = new
> > Column(column.name
> ()).setValue(column.value()).setTimestamp(column.timestamp());
> > the ctor of Column(Column ) is used, which just uses the old column.name
> > buffer, which also preserves the pos.
> >
> > is this some error on server code? (looks a flip() is needed somewhere)
> >
> > Thanks
> > Yang
>
>
>
> --
> Jonathan Ellis
> Project Chair, Apache Cassandra
> co-founder of DataStax, the source for professional Cassandra support
> http://www.datastax.com
>
