Slashdot Mirror
Debian Struggling With Security
Masq666 wrote to mention a ZDNet article discussing difficulties Debian is having with security updates. From the article: "...Lack of manpower also appears to be adding to Debian's security woes. Michael Stone, another member of Debian's security team, expressed his frustration to the organisation's security e-mail mailing list in mid-June, saying there was no effective tracking of security problems."
They have a huge team focusing on security.
Secure, Convenient, Cheap.
Pick any two.
(General rule, but it does generally follow)
I have no problem with your religion until you decide it's reason to deprive others of the truth.
$ apt-get update security-officer
Problem Solved.
(Its funny. Laugh.)
It's just a random thought, but have the Debian people ever contemplated whether their problems in this regard may stem from the fact that they have too many packages? The package list for the latest stable lists an incredible 16834 individual packages, and even though there are many programs which come in different flavours and thus contribute as more than one package, this still is a huge number.
I can certainly see why security management gets a problem here. Maybe the Debian project should cut down on these and see just how many packages are really needed.
quidquid latine dictum sit altum videtur.
Woah! Wait a moment before you start flaming me on the basis of my subject line...
The problem of providing security support is ill-suited to being solved by the traditional "mob of volunteers" approach which describes most open source development. When you're doing development, it doesn't matter if you have five people coding one week and nobody doing any coding the next week; but when it comes to dealing with a constant stream of security issues which are being reported (in particular, from upstream vendors), it is important to guarantee that there will be someone around to deal with them. When the entire security team consists of people who have other full-time jobs, it's impossible to make sure that someone will be around when they are needed.
The job of "security officer" is really one which should be a job, not a role-played-by-a-volunteer. Go out and raise some money to pay for your security officer, so that he is able to always be available when he is needed, because if he needs to get some other job to support himself, he won't be around when you need him.
Tarsnap: Online backups for the truly paranoid
Bullshit. All the technically sweet linux distributions out there which use apt are more or less resting on debian's shoulders. If you watch the security changelogs - or the regular changelogs - of ubuntu packages, you'll see that nine out of ten get made by debian, adapted to ubuntu and thrown to the ubuntu servers. Some are just renamed to "-ubuntu" and passed on. And a very few are actually maintained by ubuntu themselves.
We can't move on. Much of the linux community depends on a well-functioning debian organization. They are lacking man-power to keep their security updates as fast as the multi-employee-distributions. That doesn't mean they're technically behind, and that we have something better to move to. Although the commercial distros would love that.
Roses are #FF0000, violets are #0000FF, all my base are belong to you
It would be a hell of a lot easier if they only supported X86 architecture like all those other Distros you refer to as the ones to lag behind.
I think what they really suffer from, and I am not expert, is politics of a large system and the perception of lots of power sitting on top. I could be wrong.
Regardless of what anyone might want to say against Debian, I still believe that they are extremely good at what they do and don't get credit for it. There is no other distro out there that attempts to support as many architectures as effectively (or at all) and if Debian decided to just delete them all except X86/X86-64 then their job would be a hell of a lot easier to execute.
The article didn't go quite as in depth as I would have liked. Specifically, the Debian apt repositories have literally, and you may quote me, zillions* of packages. I'm fairly certain they have quite a few more than, say, Red Hat has binary packages in their repositories.
Therefore, it would follow that if 4% of Debian packages had security vulnerabilities that would equate to a substantially greater number of packages than would the same 4% of Red Hat packages.
The other important thing to keep in mind is that it's unlikely many users would install all zillion packages at one time.
Finally, the article implies Debian and Red Hat are in competition. However, as literate geeks will know, Debian is the OS of "Software in the Public Interest" http://www.spi-inc.org/about which is a non-profit entity. Therefore, while one could argue that Red Hat (a for-profit enterprise) and Debian are in competition for userbase, by no means are they in direct competition for 'business'.
*Debian website says "over 15490." Which begs the question, how many more than 15490? 15491?
You've got to admit there is a fundamental difference that would also cause that change of attitudes.
Debian security guys tend to have an attitude of trying to do things right. You're talking about the same people that chose to stop everything when they were compromised last year (and that was two days before a woody revision release). It's no surprise that people think of them as a good team without the necesary resources that need help. After all, they appear to do what they can with whatever resources they've got.
Microsoft, however, is known for turning a blind eye to big problems, trusting no one will find out and trying to NDA the hell out of everyone. Considering people pay big $$$ to them, and they do play dumb more often than they should, guess what the attitude toward them would be.
MS has been doing things a little better lately, but years of treating security like they did in the '90s aren't forgotten that easily.
I like Debian, and really hope they can solve their staff shortage. I wouldn't like them to go under because of this.
GPG 0x1B479C78
Supporting arches that span the gamet of bitness and endianness shakes out bugs and bad assumptions that can be hard to find otherwise. These fixes get pushed upstream whenever possible. So Debian is raising the water for a heck of a lot of boats. Until the great license blowup, Debian's X-Strike Force was also a major reason why XFree86 ran on so many platforms. The bit and endian issues THERE are a bitch.
It might be better in some respects if Debian were x86 only like everybody else but we would all be poorer for it.
"When the entire security team consists of people who have other full-time jobs, it's impossible to make sure that someone will be around when they are needed."
.002 for widget X that isn't widely used and gets abandoned for lack of interest and now has a security issue, how is that different than in the commercial world? At least with OSS someone/anyone can fix the problem. With commercial software you literally have to stop using the software because no fix will ever come.
Your wrongly basing your entire arguement on the idea that OSS programmer(s)=loner(s) with other "real" jobs. That is simply not the case for many OSS projects. Commercial OSS companies like Red Hat, Suse/Novell, et al are and have been the driving force in OSS for some time now. Look at any big distro, any major software project etc and at this point chances are they are being bankrolled and supported by commercial copanies that are paying people to work on them and deal with things like security issues. And if a popular project has a security flaw that an author won't address, and distros won't fix because its not part of their distro...well you know the deal, use the source luke.
I see what your trying to say but again your arguement is flawed as "traditional" OSS development no longer means unpaid and non-commercial. I don't think that the people buying Red Hat linux and getting security support for years and years would share the same viewpoint. And I also don't think that commercial companies put more into security than OSS programmers do. History just doesn't show that.
For version
OSS is particulary well suited to dealing with security issues IMHO and the problems it has with security are more or less the same problems that commercial software makers face. Your floating down a well known river in Egypt if you think that in the commercial world all projects have people who are paid to soley to work on security.
If you wanna get rich, you know that payback is a bitch