Re: BUG #3049: psql does not honor md5 in pg_hba.conf file

Aaron Zedonis wrote:
> The following bug has been logged online:
>
> Bug reference: 3049
> Logged by: Aaron Zedonis
> Email address: azedonis(at)ntag(dot)com
> PostgreSQL version: 8.1
> Operating system: Windows
> Description: psql does not honor md5 in pg_hba.conf file
> Details:
>
> I have set up a server on a Linux machine, in my case the IP is 10.0.0.104.
> I have configured the pga_hba.conf file so that all external requests
> require password authentication. e.g.
>
> # TYPE DATABASE USER CIDR-ADDRESS METHOD
> local all all trust
> host all all 127.0.0.1/32 md5
> host ntag ntag 10.0.0.104 255.255.255.0 md5
> host ntag report 207.145.64.220 255.255.255.248 md5
>
> When I attempt to login from a different Linux machine, I am correctly
> challenged for my password. e.g.
>
> $ /usr/local/pgsql/bin/psql -h 10.0.0.104 -U ntag ntag
> Password for user ntag:
> Welcome to psql 8.1.4, the PostgreSQL interactive terminal.
> ...
>
> When I attempt to login from a Windows machine, I am not challenged for my
> password:
>
> C:\Documents and Settings\azedonis>psql -h 10.0.0.104 -U ntag ntag
> Welcome to psql 8.1.4, the PostgreSQL interactive terminal.
> ...
>
> Logging in from the DOS prompt, or a Cygwin terminal has the same result.
>
> If (from the Windows machine) I attempt to login with the postgres user, I
> get the following error:
> psql: FATAL: no pg_hba.conf entry for host "10.0.0.60", user "postgres",
> database "ntag"
>
> So, it is reading the pg_hba.conf file.
>
> As another test, I changed the pg_hba.conf file to require md5
> authentication on all (including local) connections, and I restarted the
> postmaster. This had no impact on the above tests.
>
> The pgAdmin III tool for Windows correctly honors the pg_hba.conf file.

I think you are confusing server-side configuration with client-side
configuration. pg_hba is *only* parsed by the server, and the client has
no effect on it at all.

It looks a lot like you have a pgpass.conf file on your Windows machine,
possibly created by clicking "save password" in pgAdmin III. See
http://www.postgresql.org/docs/8.1/static/libpq-pgpass.html.

//Magnus