The large_scale_tests file is in
/afs/slac.stanford.edu/u/sf/jgates/Public/, and I attached it since it
small. About 150 lines in I put in the queries that cause problems.

The sum up so far is the simple things appear to work. Running multiple
queries at the same time is fine.

I tired a very simple join and the cluster hung. I could talk to the
czar but queries were not being answered. Canceling was not really
successful. The czar log looks like it was trying to answer the query
but nothing ever came back from the workers.

Having dec instead of decl crashed the czar.

I doesn't seem to know about  mysql.fluxToAbMag.

-John