Slashdot Mirror
New Remote Configuration App For Linux
Whyte Wolf writes: "Olympus is a new remote administration suite for Linux. Its currently in development by Mount Linux a company based out of Calgary, Alberta. This is a really neat little product, with some cool ideas (I especially like the NetMessage class) and ofcourse, it's Open Source. They're looking at developing a Windows client to allow Win32 users to administer a Linux system remotely."
-John
The obvious sarcastic comment is..."Who wants a win32 user to admin a linux system?"
:)
cheers
It's already bad enough when you edit stuff in /etc by hand and Linuxconf chokes on it. Having different admin tools disagree on the proper formatting for these files would drive the inexperienced administrator up a wall.
The veteran admin is so accustomed to using a text editor via telnet he won't consider any other option. Even in situations where it is faster.
Yes such situations do exist.
--= Isn't it surprising how badly I spell ?
This software is so far from being complete that they don't have much of anything except a few shots of mockups. They mention that it should not be used for production and that it's nowhere near complete.
This is all part of Linux's slide into the mainstream, and I'm not entirely sure that it's a step in the right direction. All the software does is put a Windows-style GUI on top of an encrypted channel - something anyone with an ssh client and UNIX competency has anyway.
In places, the 'GUI-on-top-of-CLI' is even more apparent, such as the "olympus ping" which simply opens a window with a text box and displays the result of a ping in a different format. That's very helpful, in case you forget how to spell 'ping'.
As a UNIX sysadmin, the notion that this could actually catch on frightens me. This could breed a whole new set of clueless [l]user/admins who don't know what traceroute is, or how to configure a firewall without their precious GUI. What's next, Linux Certification? Linux Certified Engineers?
The last thing the Linux world needs is abstraction of administrative functions via a GUI.
this is a sig.
It seems to me that one of the things that will forever dog linuxconf is it's inability to keep up with server changes. The manner in which they build modules has them constantly changing a rapidly moving target. It seems that this concept these folks are working on may fall into the same category.
.conf files.
.conf files would still be editable from a text editor or be got at with this somewhat more automated process. Instead of having the author of this API be responsible for keeping up to date with every possible daemon out there, it would be up to the daemon writers to provide a definition file to this API.
This is something I proposed in a Slashdot article a while ago, but it was way late in the conversation. Since this appears to be on topic again, gonna try once again to see what you guys think of it.
The concept revolves around a similar notion to linuxconf in that the idea is to provide a GUI tool for configuring network properties. Instead of trying to develop a unique graphical module for each possible daemon, build an API that looks to definition files. Each of these def files (most likely XML) would contain what text file is used to config a service, what are the possible attributes of that service, and what are the valid values of those attributes. I imagine one could also include the same kind of help text you would normally find in most
Rather than trying to reinvent the GUI for all this, instead provide this API as a means for config GUI's to be written. So whether you're writing a config file for KDE, Gnome, or just X they all call to the same API. The API would handle reading the definition files, and writing to the config files as defined.
From what I'm thinking, the
Good idea, or am I full of it?
The line must be drawn here. This far. No further.
I guess this is what Microsoft means about encouraging innovation; third-party stuff like this is actually useful on Windows. Try to sell it to Linux people, and everybody shrugs and says, "I don't need it. I've got ssh." (or whatever simple, elegant tool fits the task in question)
"The best we can hope for concerning the people at large is that they be properly armed." - Alexander Hamilton
WebMin (www.webmin.com) is a full remote configuration tool for Linux (and full UNIXes as well) via a web browser. It's cross platform, open source and much more mature IMHO. Since I started using Webmin, I never even touch Linuxconf any more. I just have no need to - WebMin does EVERYTHING.
1. We want more people/companies to use Linux as servers
2. People need to administer servers
With those assumptions in mind, let's look at the people who adminster servers. There are 4 potential categories of server admins.
1. *nix geeks
2. MC*E winders types
3. "other" (isn't Banyan Vines still in business?)
4. Newbies and aspiring server admins
Now, given assumption 1, that we want to see Linux expand, we will need to see a lot more admins be Linux savvy and Linux friendly. All of those future Linux admins will have to come from the 4 categories above. We've got category 1 wrapped up, so if we want more Linux admins it will have to be at the expense of 2, 3 and 4. Ignoring the VMS hold-overs from category 3, this boils down to acheiving 2 things:
1. Converting MC*E types to Linux
2. Convincing new admins that they would really rather do Linux than Winders
Olympus helps in both these cateogries. For the MC*E type, the ability to have access to the Linux box from a Win rig will make them more amenable to setting up a hetrogenous environment. Really, if I had a dollar for every WinAdmin who said they wouldn't consider incorporating Linux in their network because it was too tough to integrate... well, I'd have enough to go see a movie, but damn are movies expensive these days! And don't rattle off at me about Samba. Rattle off at those WinAdmins. Olympus will help gain converts from category 2. Mock MC*E's all you want... but keep in mind they can be converted and "see the light". Olympus will go a long way towards that.
Category 4, new admins, is much more fertile. Lots of aspiring admins go the MCSE route. There are a number of reasons for that, the biggest being that MCSE is an instant credetential. We have responded to that in part with the Red Hat cert. Another major reason for the continued success of NT is the illusion that it is easy to admin (it's an illusion because the troubleshooting of inevitable failures, in the long run, more than make up for the crunchy point-and-click ease of setup). By giving a straighforward config interface to Linux servers we provide newbies with real ease of use. I like to think of this in terms of "total ease of ownership". Once Linux runs, it runs. Easy. Making the config easy is step two.
2 1337 4 u!
Does it suddenly delete the entire contents of the /etc/group file for no apparent reason? Does it introduce lines consisting of nothing but commas into the /etc/aliases file? Does it put things that belong in /etc/rc.d/init.d into oddly named files in the /etc/sysconfig tree?
If it doesn't do these things, it is a VAST improvement over linuxconf. I once made the mistake of letting the sysadmins here use linuxconf to administer our linux farm, and the company barely survived.
--Charlie