Alvaro Herrera <alvherre@xxxxxxxxxxxxxx> writes: > Jeremy Finzel wrote: > >> UPDATE: what is actually failing is a call to >> pg_xact_commit_timestamp(xmin) on a given table under the view. We still >> think we must have some corruption though with pg_commit_ts. > > This is not a valid query, because the xmin may belong to a frozen > tuple. pg_commit_ts does not keep data forever; old files are removed. > Anything older than datfrozenxid could return that error. Ok but what does it mean if the error is raised only on both streaming standbys but not on master? We took a quick look to verify, just in terms of pg_commit_ts file names, between master and standbys to find no differences. Thanks -- Jerry Sievers Postgres DBA/Development Consulting e: postgres.consulting@xxxxxxxxxxx p: 312.241.7800
