I have been able to reproduce the error in my development environment. EXPLAIN output is attached. EXPLAIN ANALYZE fails with the error mentioned.
I have to do a little configuring to get the stack trace logs, but it will be a couple of weeks before they are ready. Tomorrow I head to San Francisco for a few days off prior to pgConf.
On Wed, Nov 9, 2016 at 12:50 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
Steve Randall <srandall@s3.com> writes: > running pg 9.6.1 on AWS > max_parallel_workers_per_gather = 4 > This only occurs in some schemas, not all. The affected schemas have > approx 1 million rows in the table. Much larger schemas (65 million rows) > and smaller schemas(100K rows) do not have the problem.
FWIW, I failed to duplicate this from the information provided. Can you get us EXPLAIN output for the troublesome case, or does it fail before printing the EXPLAIN output? How about a stack trace from the errfinish call?