Re: elog() patch

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > Can you take care of the echo of entered password too,
>
> I'm unconvinced that that's wrong, and will not change it without
> more discussion. (1) The reason it was put in was to allow debugging
> of "that's the wrong password" mistakes. (2) The postmaster log
> inherently contains a great deal of sensitive information, so anyone
> who runs with it world-readable has a problem already. (3) The password
> is not emitted unless the message level is a lot lower than anyone would
> routinely use. (4) If you're using the recommended MD5 encryption
> approach, then what's logged is encrypted; it seems no more dangerous
> than having encrypted passwords in pg_shadow.

OK, I have thought about how we display invalid passwords in the server
logs. This isn't an issue if the password is the same as stored in
pg_shadow. However, if the invalid password was incorrect because it
was their Unix password or a password on another machine, I think we do
have an issue storing it in the server logs. I can't think of any unix
utility that stores invalid passwords in the log, no matter what the
debugging level, and I don't think we should be doing it either.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026