Webmin’s (see my last post if you don’t know what I’m talking about) strength flows from a plethora of modules which allow you to configure and control most elements of the underlying operating system. Common applications - for example, Apache - also have officially-supported modules that ship with the main Webmin installer. In …
A few mistakes...
"If you area a systems administrator looking to deploy, or charged with maintaining, a Linux or Solaris system, see if a Webmin module exists for the services you use; it may be critical for your sanity."
However proofreading articles before you post is critical for good content...
Where you wrote:
"Sendmail’s configuration files were designed to be modified by people who think like people, and so I find the Sendmail module to be a salvation."
Did you mean "...modified by people who don't think like people..." by any chance?
My editor ate it.
It was supposed to read "Sendmail’s configuration files were designed to be modified by people who think in regular expressions."
I will go plead with him to restore it.
Isn't one of the reasons that Windows admins have problems with Linux because on Windows you have a nice GUI and it does lots of automagical stuff when you click here, here and... here? It's great for 90% of your work and you don't have to think much about what you're doing. However, it does mean you don't learn how stuff works.
Moving to Linux means you actually have to understand not only where configuration files are and their format, but also the implications of every single edit you make and how the applications/daemons you are configuring work at a much lower level. Consequently, after 10 years or so you can consider yourself no longer a newbie and start to be able to look be-sandled & bearded, grizzled old Linux gurus in the eye - at least for a second or two.
Won't Webmin reduce a lot of Linux sys admins to the level or point and click Windows admins again? Surely it risks making the situation worse? Why bother learning the ins and outs of zone files or sendmail configuration when you can just point and click?
I have spent the last few months designing the architecture for a hosted version of our software product and then implementing that architecture. Sure there have been days when I've nearly stuck my fist through my screen - and we're a little behind schedule - but I now know soooo much more than I did before. It'd be a shame to not have had that experience in my opinion...
Taking a step back
When I see a sentence like "but I now know soooo much more than I did before. It'd be a shame to not have had that experience in my opinion.", the subconscious self-justification alarm bells go off: it took me effort to learn how to do X, for which I needed to know about Y and Z, so people who want to do X should also learn Y and Z!
Or, to fill those things in: "it took me effort to learn how to drive, for which I needed to know how a transmission works, and how to work a clutch and shift gears, so people who want to learn how to drive should also learn those things". Progress passes these statements by. You want to learn to drive, you could also opt for buying a car with an automatic transmission, and never bother with a manual. Yeah, during the transitional period one was better than the other, but the idea behing progress, be it about cars or X'es, is that the only thing that counts is doing what needs to be done. The methods in which you get there changes.
Doing system administration used to require the effort to learn lots of different, often mutually inconsistent tools. Mastering them took even more effort. But the reason you learned to use them was not "to use them", it was to get something done.
So what if this metatool makes the need for the old effort go away? entirely? The tool devs keep maintaining them to do what they do, and the people who write the metatools keep making sure that their metatool talks to the tool correctly. Why would anyone not part of those two groups ever have to know how that interfacing works? Just because it's a hard thing to learn, and once you've learned it, you have something "more" in your experience toolkit, doesn't justify someone else going "yeah but.. that experience has become useless, because now, with this thing here, you don't need it ever again."
It's like the engineers of old who have learned how a flipflop works, and are then told that for practical purposes, people who actually need the functionality offered by flipflops have no intention at all in learning how they work. They just need to work. There is outrage, their knowledge is valuable, but it's also completely irrelevant.
Same thing here: you could be a sendmail god, or a httpd.conf black mage, professing how it took you years to learn everything there was to know, and how other people should try to learn half of what you have learned, but you've gone too far: I don't care about learning those individual tools, I just want to know that when I tell my tool-using-metatool to first filter mails and then land them in an inbox, that that's what it'll do. Not how sendmail does that, or how my tool tells sendmail to do it; only that it will work. The rest is not relevant to getting the job done.
Yes, if it fails, you might be stuck. "You see? you needed our help all along!". Actually, no, I really didn't. I only need help now. All those days and weeks of months when everything worked, I was able to invest my time in learning other things. Things that are hopefully going to still be valuable in ten or even twenty years. In the knowledge that just because it takes me a lot of effort to learn, someone else might breeze past me because they got a shiny new tool for the job.
"Won't Webmin reduce a lot of Linux sys admins to the level or point and click Windows admins again"
I think using the word 'reduce' there is a bit leading. I don't need a expertise in mining and metallurgy to use a spanner, provided it's a quality spanner that I can rely on and I know where to get another one.
I'm a linux administrator of 7 years experience. I have about 30 servers under my care and I configure everything via the command line because I'm not prepared to erode security, stability, portability or available resources for the sake of being able to work a little faster using graphical utilities. In my experience, the most consistent and reliable configuration utility is vi on a shell.
Bacula is testing this philosophy. It's so disproportionately complicated to configure it to do something straightforward and consistent that it is almost not worth the effort. The manual is nonlinear and inaccessible, and I have found myself learning to use bacula through the continuous wrenching of experience, instead of just reading up on it. Judging by the number of emails to the bacula users list I am not the only one. It's the one service I administer that will repeatedly behave in an unexpected manner.
If a webmin module means I can start to depend on bacula then I may have to move away from the command line on this one.
I agree with much of this thinking. Maybe "buzzinezz" people who are only interested in the short term bottom line might not like the idea of time being spent understanding stuff and really knowing what's going on. But I can't help noticing people supposedly in charge of fairly complicated systems - who are quite clueless about how they work and what to do when something goes wrong.
The debate between point-and-click interface versus command-line hard-core will probably rage on forever - but I never feel reassured when I am surrounded by people who have very little understanding of what they are doing - because they have been shielded by their favourite interface from the complicated, and time consuming detail underneath. The day when things go horribly wrong, they are useless fixtures hanging at the end of a support line, waiting for the tech gods to impart them the knowledge they have denied themselves by believing in saving time, or in using the "easy way".
"Or, to fill those things in: "it took me effort to learn how to drive, for which I needed to know how a transmission works, and how to work a clutch and shift gears, so people who want to learn how to drive should also learn those things""
I beg to differ. We are not talking about drivers, we are talking about mechanics - to use your figure of speech. An IT person is not the equivalent to a driver, he/she is the equivalent to a mechanic. A driver is the users. And indeed, I wouldn't expect a user to be clued-up to the inner workings of whatever system they are operating. And talking about auto mechanics, same principles seem to apply there. As cars get more sophisticated, a lot of them turn into button pushers, and are only good at simple "undo the bolt, change part, redo the bolt" type jobs. Look around on car related forums, and you will see stories of people taking there cars to garages with difficult or subtle faults - which keep on getting misdiagnosed and costs their owners unnecessary money. Lack of true skill and knowledge costs.
No wonder one hears so many disappointing stories about IT people. Plenty of people in technical positions, who are not technical at all. They just learned how to push a certain sequence of buttons - and have no idea what they do. Then, when things go wrong, or users ask questions just outside of their learned-by-heart sequence - their blank stair, or made-up answers shows that they are truly a driver, not a real mechanic - to continue on the analogy.
So I'm afraid the answer is - if you want to call yourself a professional, and not an amateur - you have to put your back into it and learn your stuff.
Webmin --> Webmin Configuration --> Webmin Modules
Set "Install From" to "Third party module from" and push "..."
You will be confronted with a list of registered third-party modules. There are currently two Bacula Modules. V1.3 and V1.6. Flavour text reads:
"Restore backups created using the open-source Bacula package. Supports browsing through the hierarchy of backed-up files in the database to select those to be restored. Now also supports PostgreSQL Bacula database and SQLite Bacula databases. "
I have never used Bacula, so I have no idea how good this module is. Still, perhaps worth a try for you. Hope that helps!
"Bacula is testing this philosophy. It's so disproportionately complicated to configure it to do something straightforward and consistent that it is almost not worth the effort. The manual is nonlinear and inaccessible, and I have found myself learning to use bacula through the continuous wrenching of experience, instead of just reading up on it. Judging by the number of emails to the bacula users list I am not the only one. It's the one service I administer that will repeatedly behave in an unexpected manner."
"If a webmin module means I can start to depend on bacula then I may have to move away from the command line on this one."
Sounds to me like the wrong solution. As you indirectly point out, plenty of command line tools have logical and well organised configuration options and/or configuration files. If one (or more) of them doesn't, the root problem should be fixed - i.e. the configuration file/syntax should be overhauled. There is no need to add another layer on top of the mess to try and keep it under control
On the other hand - yes, I know. Easier said then done - overhauling any complex application. But in principle, it would be fixing the problem, not plastering a massive patch on top of it. In principle at least :-)
Plenty of "amateurs" are put in a position where they are required to administer systems. Especially in SMEs. It's a rare one below about 50 people that gets a dedicated geek who actually gets paid to learn and implement nothing but IT. The dedicated sysadmins should take the time to learn, but are they to have this knowledge appear fully formed in their heads? What about admins who do know their stuff and just want the mundane stuff taken off their hands?
This is why GUIs are useful; for many people there is more in life they are called upon to do than knowing everything there is to know about the functioning of a single tool.
The modern polymath is a myth. It is not humanely possible to learn EVERYTHING there is to learn. Of course that doesn’t stop people on the internet for taking the piss out of everyone else int eh entire world for not knowing what they personally already know.
You’d think we were some sort of interdependent society of individuals somehow. One that evolved to have different individuals with different skill sets in order to ensure complete skills coverage rather than a race of gods who knew all and had access to all resources at all times.
The ones I really love are the ones who nerd rage about how everyone needs to learn the command line for every little task or they aren’t a real sysadmin. It goes completely past them that there are plenty of people in this world (especially in small business) for whom systems administration is not their day job. It might be one of literally hundreds of tasks they are expected to do. Or they may be a small business owner trying to do everything themselves to keep costs down, or any of a dozen different things.
Instead the expectation seems to be that everyone who ever touches a computer for whatever reason has the time, skill, patience and resources to learn everything there is to learn about that computer before even turning it on. Even more insane, some of these folk would have it be a legal requirement, if they could. The part where it is just a tool, no more important to 99.9% of people than a fax machine or stapler doesn’t seem to matter in the least.
I still don’t understand exactly how that demon fax machine works. Nor do I care to. I wrote a sticky note telling me which side the writing on the fax goes, and what sequence of numbers I need to press in order to get it to send. I do NOT need to dream about its bloody bytecode and issue custom patches for a small flaw in the modem firmware. I just need the bloody thing to work.
It sadly a revelation, shock, and demoralising blow to a several people when they discover that there are millions (if not billions) of people who look at computers they same way. Who cares what OS it runs, or why…so long as the tool does the job it was created for and we know how to use it, then roll on…
Oh, but that’s blasphemy to speak. Watch the thumbsdowns I get for having the termidity to mention that the emperor has no clothes…
It's really called DRBD.
I agree that webmin is a useful tool for some people, but Trevor seems to think it's the be-all and end-all of Linux, Solaris AND Windows system administration. It might lower the entry barrier for typical Windows admins, but in that case all it does is give the user a false impression of how they're able to "manage" the system.
Until something goes wrong, that is. And they have no idea how to fix it.
If one wants to go any further than this, it's imperative that they are fluent with the command line. And the only way to do that is to use it as much as possible. Learn to script things so you don't have to type as much. Learn your shell. Grep your mail logs, don't just open'em up in an editor.
Learn the inner workings of your operating system.
About 10 years ago I took the RHCE exam. I doubt I would be able to pass it if most of my daily interaction with the systems was through Webmin.
Thanks for the DRBD catch.
I do not think it's the be all and end all of systems administration. I think it's the cat's meow for newbies to the Linux or Solaris platforms. TBH, I use both SSH and Webmin for my daily workout. For some things, (such as a good YUM) the CLI is way faster.
I don’t recommend Webmin as the ONLY way to interact with a system. I think it’s a great way to LEARN a system. It’s a beautiful safety net. Once you *do* know how to solve your problems manually, (say…all your NICs going up in smoke when you clone an RHEL VM,) it’s a great way to take the administrative burden off of your daily tasks.
For me the CLI is almost a mid-game item. Webmin for the early-game…learning the ropes. CLI for the mid-game…learning what lies behind Webmin. Then Webmin again for the late game, when you just want to poke at it and have it work, in full understanding of what it’s doing and how to fix it if it breaks.
The command line is the only thing you can trust when things go bang
A long time ago it was drummed into me to get used to vi, as at the time when things went wrong it was probably the only editor that you could rely on to be around when stuck in single user trying to sort something out.
Not so long ago one of my collegues would routinely do any Vertitas work (we were running various cluster and replication products as well as the usual disk stuff) from the command line as from his past experience there are times when you're going to be stuck with the command line trying to sort something out.
I took RHCE a few years ago and a large chunk of that was to sort out a broken system from the command line as they deliberately break things like network connectivity and windowing system.
To be honest until Trevor's articles I'd never even heard of Webadmin! Though even though I now know about it, I think I'd still rather stick to getting stuff done via scripts on the command line.
You're just making it worse for yourself
If you insist on looking for magic bullet solutions like webmin (like the "Microsoft Management Console"), you're never going to learn how to admin systems well. You're coming from the Windows side of the fence and complaining that because Linux and Solaris are different, they're inferior (e.g. "Clustering systems has traditionally been a special magic power reserved for those who type arcane things into a command line and mutter to themselves a lot.") As an example, using a web-based log-viewing tool will *always* leave you missing the power of the unix command line. Does it give you regular expression searching so that you can work out which external mail systems connected to yours in a given hour? Can you track which systems are trying to brute-force attack your sshd? Etc...
As an aside, if you've picked Linux spam filtering services that are using Sendmail then you've just made things harder for yourself. Try learning more about the problem space and you'd do much better.
Webmin is no substitute for learning how it all works. It is however both a great introduction to Linux that is less scary for newbs, and it is a beautiful tool for administering a system once you DO know how it all fits together.
Webmin nice, puppet better
Webmin is a nice easy way to ease yourself into the linux pool, but sooner or later your are going to want more. Especially as the server count goes up.
So if you are looking for an easy way to manage many machines, you might want to take a look at puppet. It's not point & click, but a scripting system for systems management
check http://www.puppetlabs.com for details
It supports most if not all linux distro's & Solaris. They say on the website that support for Windows is in the works too.
@Lon John Brass
Yar. <3 Puppet. Did a different set of articles that touched on puppet a while back. Good toy, that.
Friggin titles can bite me.
Webmin has had a bad press over the years, historically because of security issues and more recently just in principle due to rabid mobs of people that tend not to manage systems but think they're awesome because they use awk.
I don't trust people who haven't managed gentoo systems. If we're really talking Linux experience of the rabid mob type your yum experience isn't going to get the job done for me.
I personally use debian for all my stuff but I've had gentoo experience, it's always felt to me like a minimum understanding level. If you can't install a stage 3 gentoo why are you applying for a job doing linux sysadmin? Nice little test that I'd prefer to any RHCE paper any day of the week. I have no trust in linux certs whatsoever.
Back to webmin, I tend to use it, just every now and then do stuff that I find tedious, I don't think I'd ever fall back on it though. If nothing else I like the efficiency of an ssh console - but I also love the efficiency of a windows gui - what it comes down to is that a web browser interface isn't really the best solution for this particular job. The StressFree theme makes it much more bearable though, highly recommend it :)
Fair enough. I'm not a dedicated Linux administrator. I have a lot of experience with Linux, but half the people in these forums could run rings around me. I'm an IT generalist. I have to maintain Windows, Linux, networks, storage, desktops, and much more. Would you trust me to run your bank's super-critical 24/7 Linux installation? You'd be a complete and utter fool to do so. I have never tried to claim otherwise. Similarly, if you have a gigantic Windows deployment and you need to eek out every single last % of efficiency and security, I am a terrible choice!
I know a great deal about all areas of IT. (Except mainframes.) I have a great deal of experience in all areas of IT (Except mainframes.) I am absolutely fantastic at innovating unique solutions to problems based on virtually non-existent budgets. I am also fantastic at research and the absorption of raw information on a given topic.
What I am not is a /specialist/ in any given area. As much as I know about any area of IT, there are a thousands or even millions of IT bodies who know more. They know more because it’s largely /all they do/. That one specific area of IT. A storage guru is going to know more about storage than me because it’s his job to know more. A true hard-core Linux admin is going to take me to school on Linux because they devote their lives to the craft.
Me? I just need to make it work. I need to know a lot about it...but not nearly as much as someone assembling high-availability clusters running millions or billions of dollars worth of software. This is why I don’t target those folks with my blog. I have absolutely nothing to offer them. Junior admins can learn from me. Similarly, folks who are asked to run an SME, or for whom systems administration is a side effect of their day jobs rather than the main focus.
There aren’t many IT generalists like me left; the requirement to specialist has killed most of us off. It won’t get me a job at a fortune 500 running their systems. It does, on the other hand, give me exposure to an absolutely huge range of products and problems which I can then write about on El Reg.
For what it's worth though...I do have a gentoo box.
Having dealt with webmin for many years, I certainly would not recommend it to anyone that isn't willing to do the hard work of fixing all its many bugs and limitations. This means you have to know how the stuff underneath actually works so you can explain that to webmin, because many of the modules in webmin are incomplete, or buggy and don't necessarily work with the version of the service you actually have installed.
If you think it will save you from insanity, you are clearly already insane.
Oh and if you do try to go fix things, welcome to inconsistent spaghetti code.
You can use it to make a nice webui for simple users, but out of the box it does not really work very well at all and is more likely to mess up your system than help you fix it.
If you are aware of any bugs in Webmin, we'd be happy to hear about them (I'm one of the developers on the project). We have an active forum, mailing list, and bug tracker. We don't ignore bug reports in any of them.
Webmin has been actively maintained for over a dozen years, and is used enthusiastically by millions of people. It is a large, complex, project of several hundred thousand lines of code, so it certainly has bugs. But, they are fixed quickly when we know about them.