Lewis, Once I figure out how to open a Jira ticket I will do so.
In the mean time I forgot to say that there is a class in Cassandra (TBaseHelper) that takes a ByteBuffer whose value is not at offset 0 and returns a ByteBuffer with the same value, but at offset 0. The fix is to zero the offset in the byte buffer with this helper class before passing the ByteBuffer to the serializer. - Chris On Sep 20, 2012, at 6:46 PM, Lewis John Mcgibbney wrote: > Hi Chris, > I can also confirm this. > It has been happening rather sporadically throughout the day and I was > actually unable to replicate or reproduce. > > This definitely merits a Jira ticket so please fire in and create one > if you have the time and are able to reproduce and describe your > experience. > > If you want a more stable Nutch/gora-cassandra/Cassandra stack please > revert to gora-cassandra v0.2 for the time being. > > Thank you > Lewis > > On Fri, Sep 21, 2012 at 12:36 AM, Chris Gerken > <[email protected]> wrote: >> Just wanted to note that I'm getting a BufferUnderflowException in >> org.apache.gora.cassandra.query.CassandraColumn in fromByteBuffer(). >> >> I've seen this problem before (see my answer for >> http://stackoverflow.com/questions/11318834/store-and-retrieve-a-float-to-from-cassandra-using-hector >> ). Basically, the value in the ByteBuffer doesn't start at offset 0 (where >> the serializer expects it to). As a result, the deserialization returns the >> wrong answer and/or the exception is thrown. >> >> I haven't had a chance to download the source, let alone build it yet, but >> in case somebody else is putting together a nightly build or something this >> would be great to include. >> >> Thanks >> >> - Chris > > > > -- > Lewis
