Glen Beane <Glen(dot)Beane(at)jax(dot)org> writes:
> I am using the copy_from command from the python psycopg2 library to do some
> bulk data loading of a postgres database. This had been working OK, until
> my script barfed because I was being careless, and it seemed to leave the
> database in a strange state that I can't recover from.
What PG version is this exactly? What does pg_stat_activity show? Is
there any indication of un-granted locks in pg_locks?
The most direct evidence about why it's stuck would probably be had by
attaching to the backend process with gdb and getting a stack trace.
regards, tom lane