[Discuss] pulse files in /tmp on RHEL 6 - followup

[Matthew Gillen]

On 03/21/2015 03:18 PM, Bill Bogstad wrote:
> The really interesting thing for me here is how our modern world of 
> PAM authentication interacts with things that I don't normally think
> requires authentication. When I saw Jerry's original note, I did some
> googling and found that this can cause problems with ssh key-only
> logins as well.   If you look at /etc/pam.d, it seems lots of
> programs use PAM for authorization/authentication and I suspect that
> there are other surprises waiting there.

I agree that this pops up more and more in places you don't expect.  And
it's not just linux.  I work on a program that uses PostGRESQL as a
backend.  Eventually someone asks to deploy on a Windows box, which I
don't have good enough reasons to tell them not to, so I tell them how.

Occasionally we run into a problem where on install of PostGresql on
windows, you have to provide a password.  After the install postgresql
is running.  However, on reboot, IF the password you gave postgresql
doesn't adhere to the password policy (not long enough, etc), then
postgresql /won't start/.  It really throws people off because it will
work until you reboot, and then never work again (until you change the
password for the windows-level postgres user to something acceptable
according to the password policy).

Matt