Re: Force pg_hba.conf user with LDAP - Mailing list pgsql-general

From Joseph Kregloh
Subject Re: Force pg_hba.conf user with LDAP
Date
Msg-id CAAW2xfchLMr6Mms1X_O3WJakE5h+NwK_Eut3UaMOLunZkNAVbQ@mail.gmail.com
Whole thread Raw
In response to Re: Force pg_hba.conf user with LDAP  (Jeff Janes <jeff.janes@gmail.com>)
List pgsql-general


On Mon, Aug 1, 2016 at 5:21 PM, Jeff Janes <jeff.janes@gmail.com> wrote:
On Mon, Aug 1, 2016 at 1:32 PM, John McKown
<john.archie.mckown@gmail.com> wrote:
> On Mon, Aug 1, 2016 at 2:49 PM, Jeff Janes <jeff.janes@gmail.com> wrote:
>>
>> On Mon, Aug 1, 2016 at 11:40 AM, Joseph Kregloh <jkregloh@sproutloud.com>
>> wrote:
>> > Hi,
>> >
>> > Is there a way to force the user being sent to LDAP?
>> >
>> > For example I have the following entry in my pg_hba.conf file:
>> > host    apdb             apuser       10.0.20.1/22           ldap
>> > ldapserver="389-ds1.sl.com:389" ldapbasedn="dc=sl,dc=com"
>> >
>> > - I will be connecting as apuser.
>> > - I will supply my own user's password.
>> >
>> > When PostgreSQL does the authentication I would like it to replace
>> > apuser
>> > with jkregloh.
>> >
>> > The reason why I want to do this is to limit power granted to a user.
>> > For
>> > example I want to be able to user my regular user jkregloh for everyday
>> > things. But when I need super user actions I will login using apuser.
>> > Now
>> > this is easy enough to do without LDAP. But if I disable my user via
>> > LDAP it
>> > would remove access from both my regular user and my superuser, that's
>> > the
>> > functionality I am looking for.
>> >
>> > I am pretty sure this is not possible, but I am floating the question
>> > anyways in hope of suggestions.
>>
>> I've wanted this as well, and for the same reason.  I think you are
>> correct, that this is not currently possible.  Only authentication
>> methods which inherently provide the authenticating user's username
>> implement the pg_ident.conf mechanism.  LDAP does not independently
>> provide a username, it only uses the one provided to it.
>>
>> I thought a quick and dirty solution would be stuff both user names
>> (the authenticating username and the database username) into the
>> existing username slot of the libpq protocol, separated by some
>> obscure character.  Then break them apart on that character, and look
>> in pg_ident.conf to make sure the specified authenticating user is
>> allowed to connect as the specified database user.  I've never gotten
>> around to implementing it, though, and I doubt it would be accepted
>> into core with the "magic character" design.
>>
>> Cheers,
>>
>> Jeff
>>
>
> Perhaps what is necessary is something akin to the UNIX "sudo" facility.
> That is, an SQL statement prefix which, if used, runs the given SQL
> statement as a PG superuser. You then GRANT(?) authority to that facility
> like you would to a table or database or ... . E.g. GRANT SUDO TO SOMEBODY;
> who could then do SUDO some other SQL statement; and that SQL statement
> would be done as if the PG user was a superuser.

You can do something like:

create user jkregloh login noinherit;
grant apuser to jkregloh;


Now once he logs in as jkrogloh he can promote himself to apuser by
using "set role apuser".  So it takes an intentional action to grant
yourself extra powers, so should be effective at avoiding mistakes.
It is not quite as emphatic as having to do an entirely separate
login, however.  Also, if you want the user to inherit from some roles
and not from others, I think you are out of luck with this approach.
Finally if you have customized user settings by "alter role apuser set
..." those will not get processed when you do a "set role apuser".


Thanks, this makes sense. It's kind of like the sudo approach mentioned earlier. They would also need to take an action to return back to their original role.

-Joseph

 
Cheers,

Jeff


--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

pgsql-general by date:

Previous
From: Branden Visser
Date:
Subject: Re: Re: Query planner using hash join when merge join seems orders of magnitude faster
Next
From: Grigory Smolkin
Date:
Subject: files in database directory