I don't think it is a type variation change. Can you provide the EXPLAIN output of your query? This looks like it is an execution plan serialization bug.
-- Jacques Nadeau CTO and Co-Founder, Dremio On Mon, Jan 4, 2016 at 8:12 AM, Andries Engelbrecht < [email protected]> wrote: > Perhaps check the data type of all the fields being used for the join. > > Select cvalue, TYPEOF(cvalue) from hdfs...... limit 10 > > and similar for tag_value on redshift. > > You can then do a predicate to find records where the data type may be > different. > where typeof(<field>) not like '<data type of field>' > > I believe there was a nice write up on they topic, but can't find it now. > > > --Andries > > > > > > On Jan 3, 2016, at 8:45 PM, Rohit Kulkarni <[email protected]> > wrote: > > > > Hello all, > > > > I am sure if not all of you, but some of you must have seen this error > some > > time - > > > > *Error: SYSTEM ERROR: IllegalStateException: Already had POJO for id > > (java.lang.Integer) > > [com.fasterxml.jackson.annotation.ObjectIdGenerator$IdKey@3372bbe8]* > > > > I am trying to do a join between Redshift (JDBC) and HDFS like this - > > > > > > > > > > *select count(*)from hdfs.drill.TAGS_US as aright join > > redshift.reports.public.us_tags as bon a.cvalue = b.tag_value;* > > > > > > I don't see anything wrong in the query. The two individual tables return > > proper data when fired a query separately. Is something missing or am I > > doing something wrong? > > > > Would very much appreciate your help! Thanks!! > > > > -- > > Warm Regards, > > Rohit Kulkarni > >
