Debian Addresses Security Problems

An anonymous reader writes "After suffering manpower shortages and other issues, Debian says it has finally addressed concerns that it was falling behind on security. Debian's elected leader Branden Robinson yesterday flagged an inquiry into the processes by which security updates are released, citing a potential lack of transparency and communication failures. It was also an appropriate time to add new members to Debian's security team, as several have been inactive for a while, Robinson said. Debian initial security problems can be found in this earlier Slashdot posting."

Comments

1000 developers?

[datadriven]

I thought debian had over 1000 developers. Don't any of them do security?

Re:1000 developers?

[RAMMS EIN]

Being able to write some software and produce packages is very different from doing security. Security is something that many, even in the developer community, don't understand, or don't understand completely. Having someone who isn't completely security savvy declare your program secure does not help you very much.

Plus, Debian likely requires a lot of security people compared to other distro's, because 1) they provide very many packages (I can't say for sure more than any other, but it's likely), and 2) they don't only fix things by upgrading packages in unstable to the latest version, but also backport fixes to the version in stable.

And in the meantime, the rest of the organization needs not to be forgotten. New packages are submitted all the time, people do like to see a new release within their lifetimes, questions have to be answered, (non-security) bugs need to be fixed, etc. etc. etc. Debian is just a huge project, and I'm impressed with how well it works.

Re:1000 developers?

[smoking2000]

Of those many developers only 5 of them where in the Security team. And of those 5 only one (Brandon) has remained active.

Due to the nature of security issues, the team had tough requirements for new members, which kept fresh blood to enter the team.

Now that this problem got the attention it unfortunatly needed, new members have stepped to the plate to strengthen the security team.

You can read more about the handling of this situation in Brandon's Project Leader Report

Re:1000 developers?

[stevey]

Branden is not a member of the Debian Security Team. (and his name is spelt with an 'e' not an 'o').

The current members are listed on the Debian Organizational chart - albeit some are less active than others.

Re:1000 developers?

[stevey]

Until recently Joey was the only active member.

In the past couple of weeks Michael Stone has become active again, which has helped.

Proof

[bondsbw]

Debian initial security problems can be found in this earlier Slashdot posting.

PROOF that Slashdot submitters have access to previous stories!

Who knew, dupes really aren't necessary after all.

Re:Can somebody tell me . . .

[Kookus]

Free software is free for you to use, not free to develope.
Software engineers need to put food on the table, so they have to get a real job when there isn't any corporate sponsorship. So now after you take out the time from their busy schedules to survive, there's not a whole lot left for a life and helping develope your free software.
Now instead of a stream-lined process where coders can churn out results, you're left with only a little bit of support from those people, sometimes they get burnt out and take a break, other times they lose all their free time and stop supporting the software. That's when you see things bog down and the need to get more people on board and all the other problems that cascade from the lack of free time.

Re:I wouldn't know

[amorformosus]

I've had it running as a webserver/nagios server for the past 3-4 months, first as Sarge was still in testing, and now as stable, and it has not failed me yet. The only time I've had to reboot or anything was when we moved the server to our new rack (not a debian issue). I've not run into any packaging problems, and as for security, it seems pretty solid.

I know it's an old discussion, but I suppose you should ask yourself what you want to run it as. As a workstation, I think sarge is a great step forward for debian; however I don't think it doesn't quite fit the needs of most workstations. But that's because it's strength is as a solid server, where updates are minimal and configuration doesn't necessarily mean a GUI.

I love debian for it's consistency and ease of configuration (once you get a feel for the way packages are configured in /etc).

I'd definitely say give it a go, if only to see the improvements from woody.

The problem with Debian

[Rosco P. Coltrane]

is that they make you jump through many loops before allowing you to help them. I have several pieces of software that I wanted to contribute to Debian, so I figured I might as well be the maintainer for them. I gave up eventually, because it's just too damn bothersome, and another Debian maintainer took my .debs over for me.

IMHO, that's why they have a shortage of manpower, because it's just not easy enough for people to jump in and help.

Re:The problem with Debian

[xmgl]

Good point. I'd agree but don't forget the fact that it is also through those rigorous processes that Debian maintains its reputation for quality.

Re:The problem with Debian

[RAMMS EIN]

Agreed. this is a problem with any large organization, and Debian is definitely one of them. These procedures exist to ensure quality, and they appear to work, but they also slow down progress. It's a double edged sword.

Re:The problem with Debian

[Phleg]

Debian has no such shortage of manpower. Doing a quick wc -l over the list of Debian developers gets 1,671 people. And that's just the development team, which doesn't include the list of Debian System Administrators (which, admittedly, is much shorter). Debian has enough people for what it does, and the list of contributors continues to grow.

The problem it was experiencing, however, was a shortage of people assigned to the security team, which has apparently now been resolved.

We need a Linux Security Information aggregator

[James Youngman]

One of the problems is that, obviously, exploits can be known by The Bad Guys but not the software maintenance community (i.e. upstream maintainer, Debian package maintainer, Debian security folk). That's obviously bad.

A less obvious but perhaps more frequent problem is where security problems are discovered and announced in upstream packages, but the information doesn't flow down to all the distributions. There's no formalised or automated mechanism by which distribution security teams get alerted to relevant upstream security fixes. You might get duscussion of the problem on a mailing list which is specific to the upstream package, but the Debian Security team can't be expected to subscribe to all those lists.

Similarly though, you can't rely on upstream maintainers reliably notifying 19 (or however many) distribution security contacts for each security-relevant release. In the specific case of Debian, this sort of thing is the Debian package maitainer's responsibility. However, there are thousands of Debian packages; some of the maintainers are very responsive and some are less so. Even the responsive ones go on vacation sometimes.

I'm an upstream maintainer. I'm pretty sure that for some of the distrubutions, nobody has subscribed to the mailing list where security problems would be announced (bug-whatever@gnu.org). In this particular exmaple, Debian isn't one of them - the Debian maintainer in this specific case is very active.

However, having a single point where Linux-relevant security announcements could go would be useful. BUGTRAQ simply isn't it (partly because its mailing list software is somewhat broken, also because of the noise level due to broken out-of-office response programs, and because solving this problem isn't the goal of that mailing list). That way, at least the Debian Security team - among others - could count on being notified reliably about known problems.

Of course then you still have a workload for the security team of analysing problems, deciding on responses and preparing NMUs. That may indeed require more people - I'm not claiming that an aggregated feed of upstream security concerns and fixes solves the whole problem.

RPM and Deb

[Anonymous Coward]

I think one of the main problems for debian stems from the use of .debs. Sure, they are still superior in a fews ways to rpms, but rpm has by and large caught up since rpm v3 and certainly rpm v4,

The baroque complexity of the debian/ subdirectory and build processes compared to an rpm .spec file is really discouraging for developers wanting to package their stuff up for debian.

Similarly, while apt trailblazed decent dependency handling, the latest versions of yum are catching up and, extremely importantly, it is far simpler to set up a yum repository than an apt one - so third party developers can very simply set up a website with a small repository and manage it themselves.

There'd be initial massive outcry I guess, but if Debian were to just adopt rpm, life would become much simpler. /usr/src/debian/RPMS ...

Re:RPM and Deb

[RAMMS EIN]

Yeah, and you had to post that as an AC just to prevent the Debian zealots (like me) from finding out your identity. :-(

I've always hated the RPM-based distros for getting more successful using an inferior technology and giving many people the impression that package management on Linux was hard, while Debian made everything easy with apt-get.

However, the times have changed. apt-get works for RPMs now, and automated package managers are finally working for RPM-based distros. Maybe the time has come for a standard in packaging land, and maybe that standard can indeed be RPM.

However, notice the many maybes. Having a standard is only helpful if every distro actually uses the same packages, and I'm not very sure that is going to happen. Without that, software still has to be packaged separately for each distribution, and there is little use for standardizing the format. In that case, the best course for Debian is to stick to their own format; if it ain't broken, don't fix it.

Re:RPM and Deb

[dozer]

I agree with half of what you say. I've made both RPMs and debs and I find that RPMs are the clear winner. They are faster to install, easier to package, and smaller. The "extra flexibility" that dpkg gives you is not only unnecessary, it's a liability.

Besides, who wants their apt-get upgrade to stop every 2 minutes and ask inane questions?? Debconf sucks! Even with priority=high it acts like a stupid nieghbor that always wants to chat. RPM gets this right: install sensible defaults and let the user change stuff using a sensible interface AFTER the package is installed.

Finally, it's looking like development on apt/dpkg is largely stalled out. At least, except for package signatures, I haven't seen a user-visible change since, oh, 2000 or so.

Yum, on the other hand... COULD IT BE ANY SLOWER?? "apt-get install nmap" takes all of 4 seconds. "yum install nmap" on FC4 takes over 30 seconds as it draws endless progress bars. I have no idea why it takes so long. I like Yum's simple config files, but it's moot until they fix its speed issue.

Connectiva got it right. It's a shame rpm-over-apt hasn't caught on.

Re:RPM and Deb

[Anonymous Coward]

Having a standard is only helpful if every distro actually uses the same packages, and I'm not very sure that is going to happen. Without that, software still has to be packaged separately for each distribution

A few conditionals in a single .spec file are often all that is needed for RedHat-Fedora-CentOS/Mandriva/SuSE . Very little effort indeed if you're depending on LSB rather than using RedHatisms.

Yes, you might still need to build different binary RPMs for the different RPM distros, but they can all come from the same source RPM.

An article here http://www.novell.com/coolsolutions/feature/11256. html goes into some depth with further references.

Re:RPM and Deb

[runswithd6s]

I think one thing people misunderstand about packages is not necessarily the format of the package itself (which is certainly important), but the robustness of the tools with which you can operate on those packages. Part of your comment is targeted in that direction, and I agree. Tools are converging in features. Improvements are being made across the board on both camps. dpkg and apt, for example, have some interesting enhancements on deck. Just check out the dpkg ChangeLog if you're looking for examples of changes already made.

Regarding format, though, I still believe DEB's win for flexibility, accessibility, and a simple, straight-forward design. Baroque is hardly the word to describe the "./debian" maintainer scripts directory. What does one find in "./debian"? control, changelog, copyright, README.Debian, rules (the build Makefile), and optionally the prerm, postrm, preinst, postinst scripts. Whatever else a maintainer puts in that directory that is useful for the build process is entirely subjective to the helper tools they might use (like debhelper).

DEB's are simply two tarballs archived together, data.tar.gz, which contains the package files themselves, and control.tar.gz, which contains the maintainer scripts. If you did not have dpkg installed on your system, but wished to extract the files and information from a DEB, you would simply use the tools ar and tar. To do the same with an RPM is to open up a hex editor to find the end of the RPM header, then use dd to cut it off and output the remaining tarball. (RPM format) How many people know or want to know how to do that?

The other flexiblity that DEB's have that RPM's don't (didn't?) is that maintainer scripts can be written in any language the maintainer wishes, as long as the interpretor is installed at the time the script runs. If you're maintaining a Perl package, it's reasonable to assume that Perl can be installed as a (pre-)dependency and used to run the maintainer scripts.

Debconf, for example, is one of those optional helper tools the maintainer is encouraged to use when questions must be asked of the user/administrator at installation time. Gone are the days when DEB's could not be installed unattended. Using Debconf allows the maintainer to provide those questions, and allows the user to view them using one of multiple interfaces, or to ignore them completely. Additionally, po-debconf makes it trivial to add multilingual support for those questions.

There is plenty of documentation, utilities, and helper tools to create a Debian package, and on-line resources such as IRC, email lists, and forums. An interesting thread to read dates back from 1996, titled "Why the .deb format?". Also, take a gander at this FAQ.

Really, comparing RPM's to DEB's is like comparing apples to oranges. RPM maintainers may baulk at the "debian/" directory and maintainer scripts, but I personally baulk at having to learn yet another spec file format for RPM's and being restricted to using librpm or a hex editor to access the data contained within the package.

Re:RPM and Deb

[myowntrueself]

RPM is superior to deb in one important way that saved my ass once.

I had managed to delete all of the symlinks under /etc Don't ask how, I just did, ok?

*Fortunately* the RPM database contained all of the information I needed to reconstruct the symlinks which were created by the packages.

I work with debian systems, so it occurred to me to see how I would achieve the same success on debian systems.

So far as I can tell, symlinks are not listed in any debian 'database' on the system where the package is installed, unlike RPM where the info is right at your fingertips.

The closest I could find for debian would be to troll through the install scripts looking for where they create symlinks.

If anyone has a one-liner which will deliver a list of symlinks that should exist on a debian system I'd like to see it. Yes, one-liner. Thats what I used on RPM.

From the RPM man page;

--dump Dump file information as follows:

path size mtime md5sum mode owner group isconfig isdoc rdev symlink

So with rpm -qa --dump

if the last field isn't an X its a symlink. Easily extracted and processed.

Ok NOW someone tell me deb is superior.

Re:Slackware -- Arch

[Rosco P. Coltrane]

Arch uses a "rolling release" schedule so use the builtin package manager to upgrade and bam! your current. The package manager even resolves dependencies!

Holy crap, I didn't realize Slack had become so modern! And just to think that I'm stuck with dpkg and apt, that can't resolve dependencies and automatically upgrade your box...

Re:Sarge

[petermgreen]

i have one server thats running sendmail rather than the debian standard exim and both aptitude dist-upgrade (the reccomended upgrade method) and apt-get dist-upgrade wanted to remove it even after i manually upgraded it to the sarge version first.

i ended up using apt-get upgrade to upgrade the bulk of the system then upgrading a load of stuff manually with apt-get install and then finally finishing the job with apt-get dist-upgrade

mind you red hat basically tell you too take the system offline and use the installer to upgrade which i find even less desirable than giving apt a bit of assistance with the upgrade process.

before upgrading read the release notes as they document other issues you could run into if you don't take care. but DO NOT follow those instructions blindly always check what apt-get or aptitude plans to remove before saying yes.

Re:I wouldn't know

[pebs]

I use slackware, myself, although I was thinking of giving Debian Sarge a try

Depends on what you're trying to achieve. If you are running a server, especially one that is exposed to the internet or a large number of users (e.g. web server), Debian stable is really great. Especially with the ability to setup automatic updates; you can set it up, and not have to really touch it for another 2-3 years.

If on the other hand you are using it for a desktop, development, or "tinkering" machine, Debian unstable or some other distro would probably be a better choice.

Re:I wouldn't know

[RAMMS EIN]

I used Slack before I switched to Debian, and never looked back. I don't know your reasons for using Slackware, but for me it was that I like to be in control and not clutter my system with useless stuff. Debian allows you a lot of flexibility, but its package management system (which I honestly believe is the best in the world) makes everything a lot easier.

You can have a very basic installation for about 100 MB. I personally think that's already a bit heavy, but it's definitely better than a lot of other distros. From there, you can get almost everything you care to mention, just by runnig apt-get install package-name. Dependencies are all taken care of automatically. You can customize how many questions you are asked during installation, from no questions to lots of options (and you can always re-run the configuration questions later).

In terms of quality, you can hardly go wrong with Debian. Everything is tested and tested again before it goes into stable (which is why there are such long times between releases), but even the packages in unstable tend to work just fine. I'd say unstable is about as up to date as Slackware-current, so if that's what you like, Debian can give it to you too.

Upgrading from one version of Debian to another is as simple as setting the right apt-repository and running apt-get update && apt-get dist-upgrade.

I don't know what more to say. Just try it for yourself.

(And for those who think I'm a Debian zealot: it's worse than that. I use OpenBSD at home. ;-) )

What I really want...

[rbochan]

... to know is:

Why the hell are slashdotters trusting news about Debian from friggen zdnet? And a blog on zdnet to boot!

I mean... c'mon... it's zdnet... with about as much credibility as The Star.

Thanks...

[rpsoucy]

Debian was my first GNU/Linux distribution. 1.3 was the stable at the time, but I ran the 2.0 unstable canidate. For a while I've used others... but I always come back to Debian. The Debian Security Team is a big part of the reason. The comunity nature of Debian, and the history of Debian represent a real important part of the Free Software comunity.

Security is often a thankless job. People only care once something goes wrong. They don't see all the work it takes to coordinate timely security responce. It should also be noted that Debian takes a proactive approach to security with the Debian Security Audit Team.

Debian lost a lot of its reputation with the delays for the current stable release. I think the future of Debian, if its to keep its reputation, will be to move to a standard release cycle of once every 2 years. Sure the Debian releases are few and far between compared to other distributions, but Debian is about software Freedom, not bleading edge technology. It provides a solid and secure OS, and most system administrators don't want to roll out a new version of an OS every 2 years, in fact, most would rather keep running an OS as long as there are security updates.

There are certainly a lot of challanges for Debian right now, hopefully the "Security Issue" goes away with this change.

Re:Thanks...

[stephenpeters]

Debian lost a lot of its reputation with the delays for the current stable release

I disagree. I run servers for commercial clients. A large number of these prefer to run some type of free software as a server platform these days. Debian is an attractive platform because of the care that goes into it. The slow release cycle means that time can be spent on thorough, careful software engineering. Distributions with faster release cycles are rarely as reliable as Debian over the longer term. I and my clients are used to spending time setting up a machine, and then leaving it in production for 4-5 years with minimal maintenance. Using Debian I have found that power and hardware failures are the main cause of unplanned system downtime.

Debian is about software Freedom, not bleading edge technology.

If you do want to use some of the newer packages from testing or unstable try using apt pinning on a stable system. Simply put apt pinning allows you to mix and match selected packages from stable testing and unstable together. A simple howto can be found here

There are certainly a lot of challanges for Debian right now

There will always be challenges for Debian. The Debian leaders seem to do just that, lead. Perhaps that is why they remain such a well regarded distribution. Do not give up on Debian because of a few negative news stories. Debian has worked well for me for years. If you stick with it it should do the same for you.

Steve

Re:Thanks...

[yack0]

People want predictability.

Sign me up for 'reliability' before 'predictability'. Not only because it's easier to spell, but for my servers that are out there, I'm not planning on that many changes.

IMHO, the stability afforded me by 'stable' is worth the occasional inconvenience of being a little behind in versions. (Or a lot behind).

Many many many people disagree with this. That is why there are other distros.

People were expecting it to come out sooner,

Why?

Who, in the Debian release process, said it would be out at a certain time? The continued party line to the question of 'when is the update going to be released' has been 'when it is ready', not 'in about X months... '. Only when it came to 'imminent release' did anyone start supplying dates.

That said, Branden was elected, I think, in no small part due to his stated commitment to more frequent releases. It is his prerogative to push for that. I certainly would welcome more frequent releases, but not at the expense of stability.

There are plenty of distros to choose from. I stick with Debian because the stability it offers. Since I have had uptimes on machines that exceed the span of release dates, I really don't mind so much. But I admit that I appear to be in the minority on that sentiment.

$.02

No burn, a reply

[rjethmal]

Debian is far from becoming irrelevant. Where did Knoppix start? Xandros? Ubuntu?

These and many other distros can be seen, under the right light, as branches on a Debian trunk. I feel fairly confident in saying that no other distro could provide a sufficiently robust and broad base upon which to build.

Ubuntu and company can do as they please. Some may, eventually, cease to be recognizable as Debian-based, but that will take a very long while.

In the meantime, Debian will continue to be an example of how large-scale projects should be run. After all, Debian has been around a long time; and in that time they have managed to build up what is arguably the largest repository of software the community has. They've also managed to support a considerable number of architectures and they've done it all quite well IMHO.

Re:GOOD

[Phillup]

The only reason to run Debian is if you believe in the politics behind the distro.

I could give a rat's ass about the politics of the distro.

Or the cost.

I run Debian because it is the easiest distro I've ever found when it comes time to update/upgrade.

I simply can't afford (nor can my customers) to take a machine to bare metal for an upgrade. And while most distros really try to make the upgrade from one version to the next easy... most are not "production quality" as far as I"m concerned.

If you want to deploy systems with a long service life, Debian is a fine choice.