set digest
omething better.
Hope that helps.
Tim Frank
>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<
On 19/03/01, 5:06:16 AM, [EMAIL PROTECTED] ("Peter Schuller")
wrote regarding Re:
pg_dumpall script that it doesn't
actually pass a username/password to psql, or if you do specify the -U
then you get an "unknown option" error when it tries to run pg_dump.
Tim Frank
>>>>>>>>>>>>>>>>>> Original Message <<
ight not be such a bad idea to
go back to the "backup" user with SELECT only permissions on the tables.
Just not sure how to set the permissions on newly created tables by
default, maybe it just has to be manually.
I would greatly appreciate comments on this idea and if it is worth
As you said, GRANT will let you do
this.
Hope that helps, and I think I got my examples right, but they were from
memory so excuse me if there is a typo or something.
Tim Frank
>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<
ying to connect. This isn't "quite" what you are
asking for, but if you make sure that users' databases are named the same
as their user account it will work. It really doesn't work for my
situation since our databases aren't specific to "users".
I was testing this on a 7.3 beta the other week to try to make it work
with LDAP authentication, and I think I only got it working if I bypased
the system-auth PAM file that everything was normally funneled through.
I don't know exactly why it wasn't working, but whenever I put a line
that use
apply a crypt patch that Bruce sent me, so I can't comment if this
work exactly the same way on the official 7.3 release. I do know I had
the same issues on the 7.3beta5 release.
Sorry for the long post, hopefully this will be of use to someone with
some better knowledge of postgres authenti
Brett,
There was a PAM thread back in early December where we discussed this.
I posted a bunch of PAM testing combinations that did/didn't work. One
of which was trying to use system accounts through "system-auth" which
is really pam_unix.so. This method would not work for me in 7.3.1 no
mat
