Live data from Hacker News

Dissecting PostgreSQL CVE-2013-1899

blog.blackwinghq.com

1–8 of 8 posts

Re: Dissecting PostgreSQL CVE-2013-1899

#2
Nice article.

> Now we have to wait for an administrator to “su – postgres”. Likely? Eh.

This isn't that unlikely. On Debian/Ubuntu, postgres is the database super user and doesn't have a password set by default so the way to administer PostgreSQL is to su to postgres and then run psql/createdb/etc (it relies on local ident authentication).

Re: Dissecting PostgreSQL CVE-2013-1899

#3
post #2

Nice article. > Now we have to wait for an administrator to “su – postgres”. Likely? Eh. This isn't that unlikely. On Debian/Ubuntu, postgres is the database super user and doesn't have a password set by default so the way to administer PostgreSQL is to su to postgres and then run psql/createdb/etc (it relies on local ident authentication).

Exactly, I have done this today!

Re: Dissecting PostgreSQL CVE-2013-1899

#6
post #2

Nice article. > Now we have to wait for an administrator to “su – postgres”. Likely? Eh. This isn't that unlikely. On Debian/Ubuntu, postgres is the database super user and doesn't have a password set by default so the way to administer PostgreSQL is to su to postgres and then run psql/createdb/etc (it relies on local ident authentication).

Yes, it's not unlikely at all. It's a timebomb just waiting to explode.

Re: Dissecting PostgreSQL CVE-2013-1899

#7

Very creative. And scary! The shell script could download and run any binary. Would be neat if there were an exploit which injected something to make further injection impossible.

ehm. anti-virus? there's a whole discussion about it, i won't go into here.

but what can an anti-virus do? best would be to patch the database (might be pretty hard). then it could write something like: echo "you have been hacked!"; exit 0 into the .profile. that will let the user spot the cracking attempt.

Re: Dissecting PostgreSQL CVE-2013-1899

#8
post #7

Very creative. And scary! The shell script could download and run any binary. Would be neat if there were an exploit which injected something to make further injection impossible.

ehm. anti-virus? there's a whole discussion about it, i won't go into here. but what can an anti-virus do? best would be to patch the database (might be pretty hard). then it could write something like: echo "you have been hacked!"; exit 0 into the .profile. that will let the user spot the cracking attempt.

What....are you talking about