Thread
-
Local Host Security? All users should have passwords optionally...
PostgreSQL Bugs List <pgsql-bugs@postgresql.org> — 2001-03-26T22:08:41Z
Ivn Baldo (ivan.baldo@pilasnet.com) reports a bug with a severity of 2 The lower the number the more severe it is. Short Description Local Host Security? All users should have passwords optionally... Long Description I wanted to add passwords to all the users on the database, including the postgres user, etc. Then everything is authenticated using "crypt" method, so it asks passwords EVERYTIME. The problem I found is that I cannot do a "pg_dumpall" anymore, since I have no way to tell it to use the "postgres" user with a given password. It tries to use the user "root" without password and it fails miserably! What happens if a hacker (or worst, a cracker!) enters to the machine somehow and I don't ask passwords for unix domain sockets? Well, it has access to all my data... Ok, this should not happen, but I worry if it happens and I think it is important to enforce the security a little more in Postgres. The documentation doesn't say anything about this... Sample Code No file was uploaded with this report
-
Re: Local Host Security? All users should have passwords optionally...
Peter Eisentraut <peter_e@gmx.net> — 2001-03-27T15:56:00Z
> Iván Baldo (ivan.baldo@pilasnet.com) reports a bug with a severity of 2 > I wanted to add passwords to all the users on the database, including > the postgres user, etc. Then everything is authenticated using "crypt" > method, so it asks passwords EVERYTIME. The problem I found is that I > cannot do a "pg_dumpall" anymore, since I have no way to tell it to > use the "postgres" user with a given password. This is a known problem. You could try to patch pg_dumpall to pass the -u option every time it calls pg_dump and psql. > It tries to use the > user "root" without password and it fails miserably! What happens if a > hacker (or worst, a cracker!) enters to the machine somehow and I > don't ask passwords for unix domain sockets? Try changing the permissions on the socket file (chmod). -- Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/