> Do you do bulk
> permissions updates that might affect thousands of role memberships at
> once?
So we do regularly update role memberships - essentially a sync from a separate database, and some could well have happened just before connections, but it's more in the tens at a time at most, not thousands... Or at least, that's what I thought. It sounds like it would be good to see if it's doing more. It'll take some time for me to figure this out though...
> I'm fairly sure that if that exists it's always noticed first,
> bypassing the need for any role membership tests. So please
> confirm whether your production database has revoked PUBLIC
> connect privilege.
I realised that in fact we hadn't revoked this. So it sounds like whatever the issue, it's not about checking if the user has the CONNECT privilege?
> It could be that the problem is not associated with the
> database's connect privilege, but with role membership lookups
> triggered by pg_hba.conf entries. Do you have any entries there
> that require testing membership (i.e. the role column is not
> "all")?
Running `select * from pg_hba_file_rules` it looks like the user column is always {all} or {rdsadmin}
Thanks,
Michal