More Info on Debian.org Security Breach

mbanck writes "James Troup (part of the Debian System administration team) has published more information on the recent compromise of four debian.org machines. The attack vector seemed to be a sniffed password of an unprivileged account, from which the attacker somehow managed to gain root and install the suckit rootkit and crack the other machines. As the machines were fairly uptodate with respect to security, an as-of-yet unknown local root exploit might be in the wild, so keep an eye on your boxen.Note that the main ftp archive running on a sparc machine was not compromised, so the exploit might not yet be ported to non-i386 architectures."

Comments

Human Error

[jefbed]

This incident reminds us of the importance of password security. It is sad to see one weak password responsible for such a breach. I think that it would be a good idea for the future to move away from the traditional unix password. An appropriate replacement would be something similar to RSA passphrase mechanism used by secure shell. A random passphrase with a minimum lenght would be idea. The user is the greatest security hole.

Re:Human Error

[Anonymous Coward]

Uhh, I dunno if you noticed, but it wasn't a password alone that did this much damage. The account broken into was unprivellaged, meaning it was just a simple user account.

In theory, a secured system can have this happen to it and the attacker will have fun deleting a single home directory before they run out of damage to do.

In practice, a single local privelage escalation attack is all it takes. Maybe this will end up being a good thing in the end, we get to find a previously unknown local root exploit, fix it and improve the Debian security practices, all in one move.

Re: Human Error

[Black Parrot]

> Random passphrase? Repeat after me: The best password is the one that isn't stikie'd to the monitor and/or keyboard.

When it comes to internet-based attacks, my yellow stickies are the securest files on my system!

Diebold, take note

[RealProgrammer]

All vendors and site administrators should take note of the openness with which the problem was dealt.

When I go to buy a car, a computer, or a stereo, and the saleslizard is cagey about any problems that come up, my trust level goes down. If they tell me all about all the problems with the thing they're selling before I even notice them, my trust level goes up. It's like a cool drink on a hot summer day.

Contrasting with Debian, how long did it take to find out that Diebold ATMs had been hit by the Nachi worm?

I'm now more inclined to trust Debian, and less inclined to trust Diebold.

One recommendation

[heironymouscoward]

Off-site logging of all accesses.

One of the first things that get wiped in an intrusion are the logs. All access logs should be copied in as near real-time as possible to a remote server that is not accessible from the machine being logged, i.e. a drop-box.

Re:In a nutshell - somehow

[Kulic]

You're absolutely right. For some reason, everyone else seems to be overlooking the fact that there is (or appears to be) an unknown root exploit out there.

Yes, you can probably guess/crack/social engineer a password if you try hard enough. That's why security is about layers, compartmentalisation and multiple types of protection, not just a single password.

If this was your box, would you be more worried that someone had managed to sniff an (unprivileged) password? Or that any one of your users can now root your box? I know which one I would lose sleep over.

Here's to hoping that the root exploit is found and patched nice and quick. Even better if it something else that's been missed and is fixed in the latest patch.

Human Error or faulty security models?

[Anonymous Coward]

SELinux would likely have prevented the root exploit from allowing this individual from doing as much harm as was done.

I think that it's time for the big names like Debian, Slackware, Red Hat etc to start implementing it on their network connected machines. It's being incorporated into the stock kernel for a reason. Use it!

What could be done better...

[rxed]

Quote: "All the compromised machines were running recent kernels[1] and were
up-to-date with almost all security updates[2]."

Well, it seems that 'almost' just isn't good enough. Perhaps there is more to the break in (like unknown holes)?

Sniffing passwords? They must be using 'almost patched' version of SSHd.

Re:#1 on Ten Immutable Laws of Security

[Gleef]

Not that I even like Microsoft's security list, since it's very Windows-centric, I'll bite.

Law #1 doesn't apply here. The intruder sniffed a password, and ran his own software. As far as I know, nobody was tricked into running malicious software. Law #1 should read, for real OS's
"Law #1: If a bad guy can persuade you to run his program on your account, its not your account anymore."

The first failure, as per this list was Law #5 "Weak passwords trump strong security." Someone didn't properly protect their password, this gave the attacker their foot in the door.

The second failure was the unidentified privilege escalation. This doesn't appear to fit any of the laws (they appear to be written assuming privilege escallation is trivial, I guess that says something about Windows). Except perhaps, Law #10: "Technology is not a panacaea". Just because we run well designed software that has few security holes doesn't mean that we run perfectly designed software that has no security holes.

Occasionally something slips through the cracks, like here, and it's good to know that real people are paying real attention, and that there are effective ways of bringing necessary systems back up in a trusted fashion. Eventually, this escallation will be found, fixed, and machines patched.

Re:So much for unbiased Slashdot

[jadavis]

Slashdotters are hypocrites and hold double-standards.

You're saying slashdot posters are inconsistant, but they're just different people who all happen to read slashdot. If you want to make a real argument, pick one person and attack their inconsistancies.

Another example is the political parties. You can't say that Democrats are inconsistant because of this, that, and the other. Democrats are a varied group, and they have many different perspectives and form their arguments in different, often contradictary ways. They just see a common means to their end, and each individual may be 100% consistant. (note: I'm not a democrat, I just used them as an example. This works with any political party that I can think of.)

Ultimately what you're doing is grouping variety of people together (slashdot readers) and then attacking the group as a whole for being inconsistant with respect to a separate issue (their perspectives about computer security).

You can do that to anyone. For example: "Blondes are so inconsistant. First they complain that the environment is being damaged, then the next week they're complaining about too much government regulation." Well, being blonde obviously has nothing to do with the topic, so of course you find inconsistancies in their viewpoint.

That type of reasoning is very simple-minded. The world is a complicated place with myriad possible groupings of people. Analogies that relate nations, corporations, SIGs, etc. to people often confuse the issue beyond repair. Microsoft isn't a "bully," it's just that the shareholders elect people that are likely to use aggressive business tactics and leverage the monopoly that they have to gain shareholder value. You can't punish MS in any way analogous to punishing a bully, because the shareholders could be long gone by now (however many years it takes to settle an antitrust lawsuit), because it's simply not a person, it's a group. Same with nations, it's a group and should not be personified. Think how much time the media has wasted talking about Bush as though he "doesn't play well with others." Nations are groups, not people.