On 19/02/17 12:03, Magnus Hagander wrote: > > > On Sun, Feb 19, 2017 at 2:01 AM, Michael Paquier > <michael.paquier@gmail.com <mailto:michael.paquier@gmail.com>> wrote: > > On Sun, Feb 19, 2017 at 9:50 AM, Michael Paquier > <michael.paquier@gmail.com <mailto:michael.paquier@gmail.com>> wrote: > > I have been poking at it, and yeah... I missed the fact that > > pg_subcription is not a view. I thought that check_conninfo was being > > called in this context only.. > > Still, storing plain passwords in system catalogs is a practice that > should be discouraged as base backup data can go over a network as > well... At least adding a note or a warning in the documentation would > be nice about the fact that any kind of security-sensitive data should > be avoided here. > > > Isn't that moving the goalposts quite a bit? We already allow passwords > in CREATE USER MAPPING without any warnings against it (in fact, we > suggest that's what you should do), which is a similar situation. Same > goes for dblink. > > If password auth is used, we have to store the password in plaintext > equivalent somewhere. Meaning it's by definition going to be exposed to > superusers and replication downstreams. Or are you suggesting a scheme > whereby you have to enter all your subscription passwords in a prompt of > some kind when starting the postmaster, to avoid it? >
The subscriptions will happily use .pgpass for example so it's not like users are forced to put password to catalog (well barring some DBaaS solutions). But I guess it would not hurt to give extra notice in docs about dangers of the various catalogs storing passwords.
I certainly wouldn't object to doing that, but if we do we should consistently do it in the other places that have work the same way (like user mappings).
