On 27/08/18 15:22, Tom Lane wrote: > Tim Clarke <tim.clarke@xxxxxxxxxxxxxxxxxxxxxx> writes: >> Last night for the first time our pg_dump backup threw this error for >> the first time. Repeated runs this morning consistently throw it too. > That's not supposed to happen. Can you create a test case, by any chance? > Presumably, it's triggered by some database schema change you made since > the last successful dump. > > regards, tom lane In answer to Adrian's question, we run: pg_dump -U (user) -C (database) then we pipe that out to gpg. pg_dump still throws the error without the pipe. I'm cutting down to find as brief a test case as possible at the moment; current hot favourite is a materialised view that's a crosstab using the functions from here https://www.postgresql.org/docs/10/static/tablefunc.html -- Tim Clarke
