Thread: is there still a memory leak with hash joins in PG 12.11 ?

On 2022-Sep-28, Zwettler Markus (OIZ) wrote:

> I found this blog post talking about a memory leak having hash joins due to a larger work_mem.
> https://gist.github.com/luhn/2b35a9b31255e3a6a2e6a06d1213dfc9

Oh dear, is that what passes for a blog post these days?

> Does anyone know if this problem is still there in PG 12.11 ?

It wasn't there in 9.5, so it seems safe to believe it isn't in 12.11.

> We are running on PG 12.11.
> We recognized an evergrowing memory usage.
> 
> work_mem = 50M
> there are some regularly hash joins.

I think it's more likely that this is a new, different bug (assuming it
is a bug at all).  But you'd have to provide more details in order to
track it down.

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/
"El Maquinismo fue proscrito so pena de cosquilleo hasta la muerte"
(Ijon Tichy en Viajes, Stanislaw Lem)