Is Webmin reliable for use in TurnKeyLinux?

Jeremiah's picture

Thanks for such a valuable product.  I really like the idea behind TKLBAM.  It's persuading me to migrate our linux servers to TurnKeyLinux in the TKL Hub.

I have a concern though.
I noticed TurnKeyLinux uses webmin.
I understand and agree with why you chose webmin instead of ebox
However, webmin was removed from Debian/Ubuntu repositories back in 2006 and the (old) Ubuntu documentation recommends not using webmin because "it is not compatible with the way that Ubuntu packages handle configuration files, and caused unexpected issues with people's systems".
Does that mean you have fixed webmin so it will work well with Ubuntu?
Can I trust that TurnKeyLinux's version of webmin will be "compatible with the way that Ubuntu packages handle configuration files" and not cause "unexpected issues with people's systems"?
If so, do you intend to submit those fixes so webmin can be used for regular Ubuntu installs?
Monte Milanuk's picture

As I read things, it got dropped from Debian primarily because the maintainer didn't have time to keep up with it, the package fell behind to the point where it was fairly pathetic and  the maintainer filed a request to have it dropped.  Once dropped from Debian, Ubuntu wouldn't touch it - which is funny in a way, because the Debian version of sage is so old its ridiculous (even in sid), but the Ubuntu maintainers won't drop it or update their version... 

I think it might be useful to read the actual comments of the former Debian package maintainer in the bug reports(1, 2)that got webmin dropped from the repositories...

It should be blindingly obvious that these packages are not really being
maintained anymore.  I have always been more or less the only person looking
after them.  I have had a lot of positive feedback from users (and even some
money on occasion) but I'm finding less and less time and motivation to keep
at it.  It is better to drop them now rather than perpetrate the cruel joke
that these are Debian-quality packages; especially because newbies often
rely on webmin to administer their systems and keep them secure.  And we owe
it to Jamie Cameron, the author of webmin, not to besmirch his name and product
with buggy crap.

...with the uploads I'm making today, I intend to orphan the following
packages as they need better maintainence than I am capable of giving
at the moment.


There's always been references to 'nasty ways that webmin messes with config files that doesn't agree with the Debian/Ubuntu way of doing business' that is used as further justification for it not being carried as a package... but that didn't stop Ubuntu from leaving the pack and pursuing the upstart system for initialization (and now Debian is finally heading the same direction, ironically) which is a lot bigger departure from the norm.  My point being... that the whole Debian/Ubuntu dependency circle can be interesting at times, and doesn't always make sense.  That said... the maintainer's comments didn't mention *anything* about problems with webmin in and of itself, but lack of available time/effort to keep up with it.

While I have seen some allegations that the changes Webmin makes to config files may cause some hiccups during dist-upgrade, I haven't managed to dig up any concrete examples of such - and most of the accusations I have found are second and third-hand, from 2005-2006 time frame.  I think the state of the art in both Debian and Ubuntu system upgrades has advanced a bit since then... but every one keeps blindly referencing the same links over and over. 

Turnkey Linux has been using it for a while now, and if it'd caused serious problems I imagine they would have gone elsewhere.  Webmin took over and maintains their own deb file that installs without too much problem - it depends on some perl-related stuff that has been obsoleted by Debian (!!!) and Ubuntu, thus has to be installed explicitly before installing webmin... but after that, it works pretty slick. 

I haven't had a system up and in place for a long period of time with custom configs and such and had to do a dist upgrade (usually scrap and rebuild after a certain point, because its a hobby for me) so *maybe* there is still something to the rumors of upgrade problems... but I'm starting to doubt it.



Jeremy Davis's picture

I haven't used Webmin extensively but am yet to run into any issues with it. IMO in comparison to eBox, TKL with Webmin is vastly superior. I had never come across these Webmin concerns before you mentioned them here (and in fact have read many 'Ubuntu server setup tutorials' that recommend it).

I too was unable to find any mention of people actually having problems with Webmin beyond (old) recommendations not to use the packages in the repo because they were so outdated. I only managed to find vague mentions of 'possible' issues (as Monte says, all dated around '05/'06). Its interesting because even the wording seems similar in the few different places that I did find these complaints. I suspect it may have only been one person's opinion that spread. I'm not sure if it was intentional but basically I think best case scenario it was well intentioned advice after a bad experience (that may not have been completely fair), worst case scenario it was pure FUD.

Jeremiah's picture

Thanks for such comprehensive and quick responses.  Of course I assumed TKL worked well with webmin but your answers have given me something more to place my confidence in.

I'm used to running servers from the command line but I'm looking forward to switching my servers to TKL and the webmin interface will be more friendly for other administrators not comfortable with Linux.

The old ubuntu documentation originally raised my concern with webmin as did not finding webmin in the ubuntu repository.  I did however find this webmin link later.

I totally agree ebox seems to be a burdensome and unnecessary way to go and I'm glad TKL used webmin instead.

Thanks for your help with this.  I look forward to more involvement with this community.

Jeremy Davis's picture

I look forward to more involvement with this community.

Welcome aboard. Always keen for more community members! :)

Liraz Siri's picture

A couple of months after we launched TurnKey it was clear that some kind of Web-based GUI would be required and we researched the alternatives. For the record, we considered Ebox but we didn't like how much bloat it would add. Webmin is much more lightweight and for its size it packs quite a punch!

We were disappointed to find Debian had dropped Webmin, but we decided we liked it so much that we would package and maintain it ourselves. Today Webmin and a large selection of Webmin modules are all available from the TurnKey package repository. As you might imagine we strongly disagree with Debian and Ubuntu on this issue.

The official reason is that Debian don't like it when one package manipulates the configuration files of another package and "messes up the upgrade process". The argument is that server admins should be required to edit all configuration files by hand in a text editor. Why? Because when you upgrade the package management system will sometimes asks you whether you want to:

  1. Keep your edited configuration files.
  2. Use the new default configuration file.
  3. Merge the two.

According to this argument Webmin is evil because if it edits these files for you, then it will be much more difficult for you to make an intelligent decision.

This argument against Webmin infuriates me because it smacks of thinly veiled elitism. It's equivalent to telling users "if you can't read a manual page and edit a text file, you have no business configuring a server!".

I made my counter-argument on ubuntu-devel when this subject came up:

Are you implying we should attempt to *force* everyone to learn configuration file formats and tweak everything by hand? You can do that if you like, Webmin doesn't change anything unless you ask it to. But for users that don't want to climb that learning curve, Webmin provides a web interface that allows users to edit configuration files in a slightly more user-friendly fashion.

If package upgrades assume users are tweaking configuration formats with Vim (my favorite) rather than a higher-level tool, perhaps we need to figure out how to fix that.

Bottom line, if an expert wants to make the argument that servers should always be maintained by experts like them they are free to do so. But forcing that view on everyone by dropping support for software that lowers the bar, in order to guarantee that you have to either hire an expert or become one in order to maintain a server - that doesn't sit well with me.

But that's what's so great about open source. If you don't agree with something you're free to go ahead and change it yourself.

Jeremiah's picture

Thanks again. So, just out of curiosity, what is the best (most reliable) way to install webmin on a standard ubuntu server?

1.  Use webmin's own deb package?

2.  Add TKL's repo and install from there? Is TKL's version specifically tailored only for TKL use?

Jeremy Davis's picture

The beauty of the TKL repo is that it contains almost alll the modules and as separate packages (so each can be installed via apt-get). Last I checked (admitedly a while ago) the official Webmin repo contained only a single big package with a default set of modules. You then need to add and remove as appropriate for starters, but again remove unrequired modules each time it upgrades.

Also the default TKL Webmin theme is much nicer than the default Webmin theme.

TBH though, why would you not be using TKL in the first place? If you want your own custom setup, then just use Core as a base and build from there!

PS I don't know about reliability, I've installed both and never had problems either way - although I have heard that Webmin has a few dependancy issues in 10.04, but they can be easily fixed by installing dependancies manually.

Post new comment