create user, user exists

If I do

alter user rpeterso with password 'thepassword';

Things start working. So it's as if the [AUTH_ARGUMENT] is being ignored.
Has this been depricated?

-Ron-

On Fri, 16 Mar 2001, Ron Peterson wrote:

From: "Ron Peterson" <ron.peterson@yellowbank.com>

I'm having a bit of authentication trouble. I'm trying to use 'crypt'
authentication. PostgreSQL 7.1beta5. My pg_hba.conf is as follows:

#local all trust
local all crypt passtest

When I do 'local all trust', everything works fine. When I invert the
comment, it doesn't.

Am I missing something?

Don't think so - I'm still on 7.1b3 and I get the same result. If I change
crypt to password everything is fine. For "local" it shouldn't make much
difference. Haven't had time to test it over a network.

- Richard Huxton

Ron Peterson <ron.peterson@yellowbank.com> writes:

I'm having a bit of authentication trouble. I'm trying to use 'crypt'
authentication. PostgreSQL 7.1beta5. My pg_hba.conf is as follows:

IIRC, you can't use crypt with a flat password file, you have to use
plain passwd authentication. (On a local connection there's not much
point in crypt anyway...)

BTW, it may help to look in the postmaster log; for many authentication
failures, the error message sent to the client is deliberately not
telling all. The message recorded in the log may have additional
details.

regards, tom lane

Ron Peterson writes:

If I do

alter user rpeterso with password 'thepassword';

Things start working. So it's as if the [AUTH_ARGUMENT] is being ignored.
Has this been depricated?

http://www.postgresql.org/devel-corner/docs/postgres/auth-methods.html#AEN13196

-Ron-

On Fri, 16 Mar 2001, Ron Peterson wrote:

I'm having a bit of authentication trouble. I'm trying to use 'crypt'
authentication. PostgreSQL 7.1beta5. My pg_hba.conf is as follows:

#local all trust
local all crypt passtest

When I do 'local all trust', everything works fine. When I invert the
comment, it doesn't.

I created file passtest using pg_passwd, with the following
ownerships/permissions. I've recreated it several times, to make sure
I didn't make a typo when I entered the password.

-rw-rw-r-- 1 postgres postgres 43 Mar 16 12:38 passtest

1006$ psql -d rpeterso -U rpeterso -W
Password:
Welcome to psql, the PostgreSQL interactive terminal.
...

(switch to 'local all crypt passtest')

rpeterso@pagoda ~
1007$ psql -d rpeterso -U rpeterso -W
Password:
psql: Password authentication failed for user 'rpeterso'

Am I missing something?

-Ron-
GPG and other info at: http://www.yellowbank.com/

--
Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/

Tom Lane wrote:

Ron Peterson <ron.peterson@yellowbank.com> writes:

I'm having a bit of authentication trouble. I'm trying to use 'crypt'
authentication. PostgreSQL 7.1beta5. My pg_hba.conf is as follows:

IIRC, you can't use crypt with a flat password file, you have to use
plain passwd authentication. (On a local connection there's not much
point in crypt anyway...)

BTW, it may help to look in the postmaster log; for many authentication
failures, the error message sent to the client is deliberately not
telling all. The message recorded in the log may have additional
details.

I misunderstood the difference between 'crypt' and 'password'. I
thought they both did a flat password file, and 'crypt' crypted the
passwords, and 'password' didn't. Instead, 'crypt' encrypts passwords
sent over the wire, and 'password' authenticates against a flat
(crypted) password file, rather than pg_shadow.

So local+crypt doesn't make a lot of sense, obviously.

So now I'm trying to decide whether I want to use 'password' or
pg_shadow for user authentication. Using 'password' seems like a broad
(and easily managed) brush, while using groups would give me a finer
degree of control over permission settings. I'm using ssl for my remote
connections, so the whole 'crypt' thing is irrelevant.

-Ron-
GPG and other info at: http://www.yellowbank.com/

Ron Peterson writes:

So now I'm trying to decide whether I want to use 'password' or
pg_shadow for user authentication. Using 'password' seems like a broad
(and easily managed) brush, while using groups would give me a finer
degree of control over permission settings.

The ability to use groups has nothing to do with the authentication method
used.

--
Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/