Re: non-superusers are allowed to drop the replication user, but are not allowed to alter or even create them, is that ok? - Mailing list pgsql-hackers

From Ashutosh Sharma
Subject Re: non-superusers are allowed to drop the replication user, but are not allowed to alter or even create them, is that ok?
Date
Msg-id CAE9k0PnN_CWBGuC5Av9HyRsi9qkbsZ29C6kG=x7vsORokNGMFA@mail.gmail.com
Whole thread Raw
In response to Re: non-superusers are allowed to drop the replication user, but are not allowed to alter or even create them, is that ok?  (Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>)
List pgsql-hackers

On Thu, Sep 30, 2021 at 7:45 PM Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com> wrote:
On Thu, Sep 30, 2021 at 3:37 PM Ashutosh Sharma <ashu.coek88@gmail.com> wrote:
>
> Hi All,
>
> While working on one of the internal projects I noticed that currently in Postgres, we do not allow normal users to alter attributes of the replication user. However we do allow normal users to drop replication users or to even rename it using the alter command. Is that behaviour ok? If yes, can someone please help me understand how and why this is okay.
>
> Here is an example illustrating this behaviour:
>
> supusr@postgres=# create user repusr with password 'repusr' replication;
> CREATE ROLE
>
> supusr@postgres=# create user nonsu with password 'nonsu' createrole createdb;
> CREATE ROLE
>
> supusr@postgres=# \c postgres nonsu;
> You are now connected to database "postgres" as user "nonsu".
>
> nonsu@postgres=> alter user repusr nocreatedb;
> ERROR:  42501: must be superuser to alter replication roles or change replication attribute
>
> nonsu@postgres=> alter user repusr rename to refusr;
> ALTER ROLE
>
> nonsu@postgres=> drop user refusr;
> DROP ROLE
>
> nonsu@postgres=> create user repusr2 with password 'repusr2' replication;
> ERROR:  42501: must be superuser to create replication users

I think having createrole for a non-super allows them to rename/drop a
user with a replication role. Because renaming/creating/dropping roles
is what createrole/nocreaterole is meant for.

Well, if we go by this theory then the CREATE ROLE command shouldn't have failed, right?

--
With Regards,
Ashutosh Sharma.

pgsql-hackers by date:

Previous
From: Ashutosh Sharma
Date:
Subject: Re: non-superusers are allowed to drop the replication user, but are not allowed to alter or even create them, is that ok?
Next
From: Amit Kapila
Date:
Subject: Re: pgsql: Document XLOG_INCLUDE_XID a little better