On Tue, Aug 26, 2025 at 9:01 AM Dominique Devienne <ddevienne@gmail.com> wrote:
On Tue, Aug 26, 2025 at 2:54 PM Ron Johnson <ronljohnsonjr@gmail.com> wrote: > On Tue, Aug 26, 2025 at 3:01 AM Dominique Devienne <ddevienne@gmail.com> wrote: >> ERROR: permission denied: "RI_ConstraintTrigger_c_1226298044" is a >> (yes, that's a large OID... For a 1 year old DB) > > PG's OID allocation of "user-land" OIDs doesn't start at 16384 anymore.
This is a 17 cluster. So not even 1 year old then. So getting to 1.2B OIDs in 10 months, doesn't bode well for the longevity of that cluster. But we're getting OT here. This is not a "production" server, lots of CIs and manual testing there.
Like I said, it's not completely linear. The real question, though, is whether PG looks for gaps in oid allocation once it wants to try and allocate an oid of uint32 max.
I bet it does. If it doesn't... pg_dump/pg_restore, baby!
