Will be more difficult than supporting Debian, but given the amount of interest expressed this deserves serious thought. In theory, the hard part isn't creating the appliances on any given distro. That's an infrastructure problem we can solve. The hard part will be all the additional testing required. But maybe in practice this won't be such a big deal.
Doesn't lead to as much rapid deployment & makes any project MUCH more difficult to maintain..... The system performance is always lower (sometimes only slight bits) as well. Not to mention the fact that CentOS is falling waaaayyyy behind already. Don't worry about RPMS & keep up the great .deb focussed work you all have been doing! :D
If it doesn't take away from the superior turnkey products you all have produced, I'd kind of like to see a CentOS version of things and here's why. In the industry, Fortune 500 companies predominantly use RedHat. Why? Support. Redhat Support contracts will also cover CentOS & Scientific builds, sometimes even specified in the contract.
I'd be willing to put in some hours helping to test & bugcheck Scientific or CentOS builds.
I use Turnkey almost exclusively as VM's in VirtualBox, running on Ubuntu LTS hosts... so as far as that goes, I just want OVF and virtual kernal tweaks supported well... I wont mind much if it switches to Debian other than having to learn their little differences... I am more interested in appliances being improved and rapid developement of new appliances... as long as the base OS is LTS, or wont be outdated and out of support in 6 months, I am happy.
What reasons has TurnKey decided to consider moving to Debian from Ubuntu? Debian is great, but let's not throw the baby out with the bath water... there is a reason Ubuntu is so successful. What will switching to Debian do for developers, what will it do for the end users? What will change? I don't really know, someone tell us? Turnkey let's me spend less time admining... and LTS with easy security updates is clutch... that's why I use it.
Also, thanks for all the work you guys do on TurnKey, top class stuff!
Debian is reknowned for being rock solid stable and secure. Ubuntu makes some concession to userbility and newer versions of software. Debian has no commercial backer so support contracts for (which large enterprise often requires) is not easily available (although I would imagine that the TKL devs would offer that if they released Debian based appliances). Ubuntu OTOH has Cannonical. Debian has a huge user and dev base and has heaps more packages in it's repos. They both have pros and cons. So the TKL devs are not planning to move away from Ubuntu, just include the option of Debian based appliances as well.
AFAIK at this stage it is not completely clear whether all the exact same appliances will be available on both platforms. For example I have suggested that the Debain based appliances be setup more for security and stability; keeping true to Debian. The tradeoff for that though is usually older versions of software (with all security patches backported to the older version and supplied via the repo).
Ubuntu does this too, although many of the TKL appliances include newer versions of software installed from upstream. The advantage is that new features are included and often bugs are fixed (some non security bugfixes are never backported to repo packages). The downside is that security patches (to the app installed from upstream) are not auto applied.
From a development standpoint, due to their similarities it is relatively easy to offer Debian and Ubuntu based appliances side-by-side. Many of the TKL under the hood customisations and software will run on either with little or no tweaking.
It's true that limited resources do require us to prioritize what features we invest in first. However in this particular case, 64-bit and Debian support require the same or very similar upgrades to our development infrastructure, which is what we're working on.
Ubuntu may be fine for some, so I second the idea of a generic build framework to base on any distro (What about integrating ultra fast, ultra tiny distros such as Puppy?) In any case, I wouldn't like to have to be Ubuntu based. Last time I accepted a major Ubuntu update, it broke my carefully tweaked client settings by pushing a major release of Firefox, and that broke a lot! I don't want to have to deal with that sort of stuff, anywhere, ever again!
In my experience, despite the fact that they have the same base, Ubuntu desktops vs Ubuntu servers are quite different beasts, and how I treat them is also quite different. With desktops I tend to do full updates fairly frequently and often within the early days of a release, these updates come thick and fast and can sometimes have not so great side effects. With servers I tend to very rarely apply updates (other than security ones - which are auto-applied in TKL anyway). This means the risk of updates doing anything untoward are almost nil.
I agree that more recent releases of Ubuntu are becoming more bloated but with most things there is always a trade-off. In my mind Ubuntu offers many things that just aren't available with other Linux OS - such as the breadth of pre-packaged software, official, but also via 3rd party repos and PPAs. It also has copious amounts of online documentation and other support materials.
I have tried Debian, CentOS and Fedora, as well as some of the 'micro distros' (such as Puppy and DSL). Whilst they all have positives and are quite useable, for one reason or another I keep coming back to Ubuntu and/or Ubuntu based distros (such as TKL) for the above mentioned reasons. Also many corporate operations require comercial support contracts with their server software and Canonical (with Ubuntu) is one of the only ones I know that offer this (other than RedHat).
So I think it'd be great to offer other alternatives (such as Debian or even CentOS), but to move away from Ubuntu altogether would be a step in the wrong direction.
Linux is linux. I know that people have different userland choices, but once you have grasp of Debian, Ubuntu isn't far off, neither is mint. Once you know Debian, moving to Gentoo and / or Slackware is pretty easy too - they all have the same filesystem layout. Switching to fedora/rhel/centos is different, but once you figure out their init systems... it's simple. I think the O/S choice *should* be up to the people producing the product. If Joe Blow complains because it's not his preferred O/S, well, tell him to stuff it and maybe if that person is motivated for more than just complaining about stuff, he'll make a new turnkey idea based on his/her O/S. I'm a freebsd / solaris guy. Don't actually like linux much, but it's simple and does the job. these TK images are just the ticket. The fact they run on my least liked version of linux matters little. The fact that I didn't put in the effort to get ProjectPier up and running does. Your images are perfect, just keep it up with the perfection, you have it now. Make the 64 bit stuff, stay within your aread of comfort and you'll keep producing a product that many many will use.
I think this is the best way to describe why TurnKey is actually good as it is. This is about having a high-quality turn-key solution to the most wanted features that free software can offer, regardless the foundation.
The comments on this page don't mention issues with the current TurnKey core and appliances, which is actually a good thing. So we need a 64-bit version for people using TKL on servers with more than 4GB of RAM, then we need some more useful appliances, testing and maintenance to what's been released.
It's easy to want something but a lot harder to get it done so people should simply understand its in everyone's best interest if the TKL developers focus on what's really important than splitting their effort on more than they can cover. Unless the TurnKey Linux users provide regular donations to support new developments and maintenance we should simply trust and respect the TKL team to do what they are already good at.
Guys, this is a great job you're doing. Thank you!
Dont know if this is the right location for a request but here it goes, it would be really nice to have an apliance with LDAP integrated hope some day this can be done.
It's getting a bit dated now, but should still work although may need a little tweaking? Have a look here. If you try it out, be great if you could provide some feedback (on that thread).
Please go for Debian stable 64bit
- Better for servers than Ubuntu (more stable packages)
- netinstall iso allows minimal base systems
- Not affected by corporate strategies
- Official Debian backports and Debian multimedia
provide moreup-to-date packages if needed
CentOS: Couldn't care less
debian, slackware, archlinux and gentoo would be 4 good options. any option is better than ubuntu, of course. ubuntu is good for some feature demoing to get people into linux but nothing more. it's way too unstable and unreliable for daily use. i've seen way to many people getting burned by ubuntu. only fools use it daily nowadays.
But perhaps that's true of the non-LTS releases. Generally it's pretty good on the LTS releases (which is what TKL is based on) in my experience anyway...
And I think we may see TKL Debain based appliances but I don't think we'll see any of the others anytime soon, as all the core TKL development and build infrastructure is Ubuntu based. That makes it pretty easy to support Debian as well but not so much with other Linux distros.
And I don't think the core devs have any plan to move away from Ubuntu. Not any time soon anyway.
You may be tempted to compare the two editions but there's no need to use a production server as a desktop machine by installing the graphics stack and all the "fat" that goes with it. And this goes the same for any server setup regardless the distro.
Sure a beginner (with servers) is tempted to avoid the console and exclusively use graphical tools with a mouse but as long as the server can be managed from the browser this wish is already granted to these users. But if a beginner actually wants the graphics server and probably remote desktop on the server then I see something wrong here. Not because managing servers should be restricted to the console but to skilled people who know exactly what they're doing. Because a live production server you can trust should be managed by a knowledgeable system administrator. Such a guy will never be afraid of a web interface nor the console but he will gladly use them because (s)he will save hardware resources which are not unlimited.
Just consider the stress on a server with 10 or 20 virtual machines on it, where each machine must pull off update packages for every installed package. Consider the network stress, the disk space involved, the added complexity and greater chance of something going wrong in any customized environment and so on.
I'm not saying it's the best, but from my experience I can tell you that Ubuntu Server is a good choice that I used the past 3-4 years without any breakage. And I'm not a professional server administrator but rather a hobbyist but I was willing to learn what I had so I can keep my (live, production) virtual machines healthy and secured with the least amount of software installed on those machines. I won't say it's easy, especially when you have a busy life outside this hobby, but I gained new skills, some good servers and that experience I could use to help others and earn some money for it. So it paid off pretty well.
I think it would be a mistake to try to support several different distributions as the base. It would just increase the amount of testing required and complicate things.
Isn't the whole point of this that the server is plug-and-play, without having to install and configure the applications? If it's sort of a black box to perform a service, who cares what's inside as long as it's stable and gets the job done?
Comments
Debian is fine
I would prefer a CentOS based appliance. Or even a Fedora based one.
--
"Using Linux since November 1991"
http://ohai.im/Nilrin
It will be a cold day to see
It will be a cold day to see RPM based appliances. I wish the Asterisk project could move away from there.
i would like Debian or CentOS
i would like Debian or CentOS
I agree
It would be nice if we could choose Ubuntu or a Centos base for each appliance.
Or
BSD...
A thought:
- set a meta standard for the base os
- a workflow to set up any basic appliance
should be trivial to create turnkeys with any distro.
just a thought. more thought needed. ;)
OpenBSD for security and
OpenBSD for security and FreeBSD for its stability and NetBSD to run on your toaster!
Thoughts on supporting CentOS
RPM, though handy & in use, is inferior
Doesn't lead to as much rapid deployment & makes any project MUCH more difficult to maintain..... The system performance is always lower (sometimes only slight bits) as well. Not to mention the fact that CentOS is falling waaaayyyy behind already. Don't worry about RPMS & keep up the great .deb focussed work you all have been doing! :D
If it doesn't detract.....
If it doesn't take away from the superior turnkey products you all have produced, I'd kind of like to see a CentOS version of things and here's why. In the industry, Fortune 500 companies predominantly use RedHat. Why? Support. Redhat Support contracts will also cover CentOS & Scientific builds, sometimes even specified in the contract.
I'd be willing to put in some hours helping to test & bugcheck Scientific or CentOS builds.
gentoo for me
I would like gentoo stage3 so I can figure out after this what I need myself.
Dont Bother
As fas as the product is bootable and in .iso image and works well with oracle VM VirtualBox , I am happy. :-)
Wether debian , ubuntu , fedora , unix , linux and what not.... if they run on VirtuaBox... I am happy...
CentOS 1 vote
Why...I put security first, then simplicity-maintainabilty. Appearance-usability last but certainly not least. :)
Ubuntu and VirtualBox
I use Turnkey almost exclusively as VM's in VirtualBox, running on Ubuntu LTS hosts... so as far as that goes, I just want OVF and virtual kernal tweaks supported well... I wont mind much if it switches to Debian other than having to learn their little differences... I am more interested in appliances being improved and rapid developement of new appliances... as long as the base OS is LTS, or wont be outdated and out of support in 6 months, I am happy.
What reasons has TurnKey decided to consider moving to Debian from Ubuntu? Debian is great, but let's not throw the baby out with the bath water... there is a reason Ubuntu is so successful. What will switching to Debian do for developers, what will it do for the end users? What will change? I don't really know, someone tell us? Turnkey let's me spend less time admining... and LTS with easy security updates is clutch... that's why I use it.
Also, thanks for all the work you guys do on TurnKey, top class stuff!
It's not either/or, it's both!
Debian is reknowned for being rock solid stable and secure. Ubuntu makes some concession to userbility and newer versions of software. Debian has no commercial backer so support contracts for (which large enterprise often requires) is not easily available (although I would imagine that the TKL devs would offer that if they released Debian based appliances). Ubuntu OTOH has Cannonical. Debian has a huge user and dev base and has heaps more packages in it's repos. They both have pros and cons. So the TKL devs are not planning to move away from Ubuntu, just include the option of Debian based appliances as well.
AFAIK at this stage it is not completely clear whether all the exact same appliances will be available on both platforms. For example I have suggested that the Debain based appliances be setup more for security and stability; keeping true to Debian. The tradeoff for that though is usually older versions of software (with all security patches backported to the older version and supplied via the repo).
Ubuntu does this too, although many of the TKL appliances include newer versions of software installed from upstream. The advantage is that new features are included and often bugs are fixed (some non security bugfixes are never backported to repo packages). The downside is that security patches (to the app installed from upstream) are not auto applied.
From a development standpoint, due to their similarities it is relatively easy to offer Debian and Ubuntu based appliances side-by-side. Many of the TKL under the hood customisations and software will run on either with little or no tweaking.
Cut the bullshit already. give us 64bit ubuntu!
Instead of wasting time on other nonsense.
Actually both depend on the same infrastructure upgrades
Opting out of bloated Ubuntu distros; Micro distros too (Puppy?)
Ubuntu may be fine for some, so I second the idea of a generic build framework to base on any distro (What about integrating ultra fast, ultra tiny distros such as Puppy?) In any case, I wouldn't like to have to be Ubuntu based. Last time I accepted a major Ubuntu update, it broke my carefully tweaked client settings by pushing a major release of Firefox, and that broke a lot! I don't want to have to deal with that sort of stuff, anywhere, ever again!
Server & Desktop, very different...
In my experience, despite the fact that they have the same base, Ubuntu desktops vs Ubuntu servers are quite different beasts, and how I treat them is also quite different. With desktops I tend to do full updates fairly frequently and often within the early days of a release, these updates come thick and fast and can sometimes have not so great side effects. With servers I tend to very rarely apply updates (other than security ones - which are auto-applied in TKL anyway). This means the risk of updates doing anything untoward are almost nil.
I agree that more recent releases of Ubuntu are becoming more bloated but with most things there is always a trade-off. In my mind Ubuntu offers many things that just aren't available with other Linux OS - such as the breadth of pre-packaged software, official, but also via 3rd party repos and PPAs. It also has copious amounts of online documentation and other support materials.
I have tried Debian, CentOS and Fedora, as well as some of the 'micro distros' (such as Puppy and DSL). Whilst they all have positives and are quite useable, for one reason or another I keep coming back to Ubuntu and/or Ubuntu based distros (such as TKL) for the above mentioned reasons. Also many corporate operations require comercial support contracts with their server software and Canonical (with Ubuntu) is one of the only ones I know that offer this (other than RedHat).
So I think it'd be great to offer other alternatives (such as Debian or even CentOS), but to move away from Ubuntu altogether would be a step in the wrong direction.
Please no CentOS
Please don't get distracted by CentOS. Debian is better.
Seriously, this is an issue?
Linux is linux. I know that people have different userland choices, but once you have grasp of Debian, Ubuntu isn't far off, neither is mint. Once you know Debian, moving to Gentoo and / or Slackware is pretty easy too - they all have the same filesystem layout. Switching to fedora/rhel/centos is different, but once you figure out their init systems... it's simple. I think the O/S choice *should* be up to the people producing the product. If Joe Blow complains because it's not his preferred O/S, well, tell him to stuff it and maybe if that person is motivated for more than just complaining about stuff, he'll make a new turnkey idea based on his/her O/S. I'm a freebsd / solaris guy. Don't actually like linux much, but it's simple and does the job. these TK images are just the ticket. The fact they run on my least liked version of linux matters little. The fact that I didn't put in the effort to get ProjectPier up and running does. Your images are perfect, just keep it up with the perfection, you have it now. Make the 64 bit stuff, stay within your aread of comfort and you'll keep producing a product that many many will use.
Mature thinking
I think this is the best way to describe why TurnKey is actually good as it is. This is about having a high-quality turn-key solution to the most wanted features that free software can offer, regardless the foundation.
The comments on this page don't mention issues with the current TurnKey core and appliances, which is actually a good thing. So we need a 64-bit version for people using TKL on servers with more than 4GB of RAM, then we need some more useful appliances, testing and maintenance to what's been released.
It's easy to want something but a lot harder to get it done so people should simply understand its in everyone's best interest if the TKL developers focus on what's really important than splitting their effort on more than they can cover. Unless the TurnKey Linux users provide regular donations to support new developments and maintenance we should simply trust and respect the TKL team to do what they are already good at.
Guys, this is a great job you're doing. Thank you!
Apliance Sugestion
Hello
Dont know if this is the right location for a request but here it goes, it would be really nice to have an apliance with LDAP integrated hope some day this can be done.
Thank you !
Adrian did a patch for that
It's getting a bit dated now, but should still work although may need a little tweaking? Have a look here. If you try it out, be great if you could provide some feedback (on that thread).
Go for Debian stable 64bit
debian, slackware, archlinux
debian, slackware, archlinux and gentoo would be 4 good options. any option is better than ubuntu, of course. ubuntu is good for some feature demoing to get people into linux but nothing more. it's way too unstable and unreliable for daily use. i've seen way to many people getting burned by ubuntu. only fools use it daily nowadays.
That's a bit harsh!
But perhaps that's true of the non-LTS releases. Generally it's pretty good on the LTS releases (which is what TKL is based on) in my experience anyway...
And I think we may see TKL Debain based appliances but I don't think we'll see any of the others anytime soon, as all the core TKL development and build infrastructure is Ubuntu based. That makes it pretty easy to support Debian as well but not so much with other Linux distros.
And I don't think the core devs have any plan to move away from Ubuntu. Not any time soon anyway.
The server and desktop editions are different
You may be tempted to compare the two editions but there's no need to use a production server as a desktop machine by installing the graphics stack and all the "fat" that goes with it. And this goes the same for any server setup regardless the distro.
Sure a beginner (with servers) is tempted to avoid the console and exclusively use graphical tools with a mouse but as long as the server can be managed from the browser this wish is already granted to these users. But if a beginner actually wants the graphics server and probably remote desktop on the server then I see something wrong here. Not because managing servers should be restricted to the console but to skilled people who know exactly what they're doing. Because a live production server you can trust should be managed by a knowledgeable system administrator. Such a guy will never be afraid of a web interface nor the console but he will gladly use them because (s)he will save hardware resources which are not unlimited.
Just consider the stress on a server with 10 or 20 virtual machines on it, where each machine must pull off update packages for every installed package. Consider the network stress, the disk space involved, the added complexity and greater chance of something going wrong in any customized environment and so on.
I'm not saying it's the best, but from my experience I can tell you that Ubuntu Server is a good choice that I used the past 3-4 years without any breakage. And I'm not a professional server administrator but rather a hobbyist but I was willing to learn what I had so I can keep my (live, production) virtual machines healthy and secured with the least amount of software installed on those machines. I won't say it's easy, especially when you have a busy life outside this hobby, but I gained new skills, some good servers and that experience I could use to help others and earn some money for it. So it paid off pretty well.
Don't multiply base distributions
I think it would be a mistake to try to support several different distributions as the base. It would just increase the amount of testing required and complicate things.
Isn't the whole point of this that the server is plug-and-play, without having to install and configure the applications? If it's sort of a black box to perform a service, who cares what's inside as long as it's stable and gets the job done?
Other options
http://en.wikipedia.org/wiki/Plan_9_from_Bell_Labs
or
http://en.wikipedia.org/wiki/Inferno_(operating_system)
or
FreeBSD...
or
Just keep Ubuntu!
Post new comment