Anyone forking CentOS6?
CentOS6 will continue to be supported until the year 2020, which is quickly approaching. CentOS7 uses systemd.
Are there any efforts in the works to fork a new distribution from CentOS6 and upgrade its packages to more recent versions, as a systemd-free CentOS7 alternative? I keep expecting to hear of it happening. Since Debian had enough users to fork off Devuan, there should be plenty enough CentOS/RHEL users to do the same. But the years tick by and I hear nothing. So I have to ask, is anyone out there even thinking of doing it? Hardly a day goes by without someone in my extended social circle complaining about systemd breaking something or another. Today one was complaining about systemd-logind mishandling a SIGWINCH and segfaulting, and another was complaining about systemd double-mounting a mysterious filesystem which appeared in no configuration files and spewing errors into the ring buffer. Several folks at different companies have reported that systemd systems almost never shut down cleanly -- they get to a point in the shutdown process and hang. Continuing to use CentOS6 seems like the obvious solution, but since the CentOS project will drop CentOS6 that will not be an option much longer. Has anyone heard anything along these lines? |
Not likely. The raison d'etre for CentOS is to be a binary compile of RHEL. CentOS6 will likely go away when RHEL6 does.
Note that the CentOS project is actually now supported (as a project) by RedHat itself. RHEL6/CentOS6 is based on point in time from older upstream packages for most things from kernel on down (i.e. they still use 2.6.x kernel even though kernel 4.x is out in the world and RHEL7/CentOS7 uses 3.x). This means to fork CentOS6 you'd have to do what RedHat currently does with RHEL6 (that then gets propagated to CentOS6) which is commit to backporting security and bug fixes from newer upstream packages into the older point in time packages. This is quite an undertaking and really only done because RedHat does that with RHEL because many businesses don't want to do rapid change. You might find other rpm based distros that haven't gone down the systemd path but my own take is almost everyone is doing it so resistance is futile - you WILL be assimilated. |
Quote:
To make such a fork a feasible alternative to RHEL7/CentOS7 it would need to clone its own yum repository, upgrade some of its packages (including the kernel) and then commit to testing those packages and backporting bugfixes and security patches to those packages, as well as to packages retained from CentOS6. As long as it remained familiar to those currently working with CentOS6 and refrained from breaking anything (which I know is a tall order), it seems like it would have its niche of grateful users. |
Anyone that needs to stay on RHEL6/CentOS6 will simply do so without support/updates.
There isn't really much reason to fork CentOS6 once RHEL6 is gone. It much better to plan on migrating on to RHEL7/Centos7 (or RHEL8/CentOS8 which should be released before they kill RHEL6/CentOS6 - RedHat supports the last 2 "major releases" so would need a RHEL8 before they kill RHEL6). It seems more likely someone would do a fork of an rpm release that uses newer upstream packages without using systemd than that they'd fork an old rpm package like CentOS6. However, you can be the "somebody" that does the fork of CentOS6 if you see value in doing that. The problem with extended support is people never want to move on - we're forcing folks here to go to RHEL6 (at least) and preferably RHEL7 because RHEL5 is finally at its end of support life after 10 years. Personally I've not seen much issues with systemd other than the learning curve and I often suspect most of the resistance to it no matter how folks present their objections is more "we don't like it because it is different and we don't want to learn a new way" than any real technical issues. I've seen such resistance to new things many times in my life and somehow none of the dire consequences predicted for the newfangled ways ever come to pass. There are of course technical issues with any new implementation but that doesn't mean there weren't also technical issues with the old way of doing things either. There are daily patches being released to many pieces of code that have been in the wild for years. |
Talk of systemd reminds me of something I'd posted to another list back in 2015:
Quote:
|
Furthermore, christian scientists now have proof that both L. Poettering and L. Torvalds are incarnations of the antichrist.
|
Please, folks :-) let's not turn this into "one of those" systemd threads the moderators despise so much.
The question is mainly whether anyone was aware of a project to fork RHEL 6 and/or CentOS 6 into a distribution which would be supported beyond 2020. I'd like to leave the thread up for a little while to give folks the opportunity to chime in. If nobody speaks up by the end of May, I'll mark the thread "SOLVED". |
And let this be the first and last official warning. Please keep on topic or the thread will be closed.
|
I'll agree that it makes no sense to try to take Centos7 and make such a major change. It may be possible to do it but more trouble than it's worth. Almost every common distro has the ability to be edited to use almost anything going. You'd have to spend forever trying to undo the very things you want out of a distro. Just get a distro that suits you or you'll have to build your own. Plenty of folks use LFS here.
This web page has a great list of choices. Don't see any RH based. http://without-systemd.org/wiki/index.php/Main_Page |
Quote:
Quote:
|
It's not going to happen. Apart from the difficulty, where's the demand? Most users of CentOS are corporate or institutional: internet companies, computer companies, or universities who want an enterprise Linux but don't need support. They won't be interested in trusting to something cooked up by a bunch of amateurs, to put it bluntly. Home users of CentOS who don't want systemd are a minority of a minority.
I'm not surprised that some people have experienced problems with systemd: it's new. I dare say there were plenty of problems when Linux was new, but that's before I arrived! |
I expect there would be enough demand among precisely the market you describe -- "internet companies, computer companies, or universities".
At my own employer (an internet company) the IT department kicked tires on a lot of alternatives (including FreeBSD) before reluctantly switching to CentOS7. They were looking for a systemd-free alternative. Now that CentOS7 is in use, it is reviled and openly cursed on a daily basis. It has caused us no end of problems (failures to log events, device mismanagement, inappropriate filesystem automounting, overwritten network control files in /etc, mishandled signals, unclean shutdown). My sysadmin friends at other technology companies have had similar experiences, and would love to be using something else (even if it's just CentOS6), but they feel they have no choice. If IT departments had a modern RHEL-like option unburdened by systemd, I think some of them would take it. If Ubuntu can gain traction as a server OS (and it has), then anything can, whether it's "something cooked up by a bunch of amateurs" or not. So far I've found one guy seriously digging into forking CentOS. He's made a good start setting up a CI system for it (just straight CentOS 6.5 so far). I've asked him to write a little about his project but so far he's only said "too busy, ask later". Time will tell if that pans out. As for this thread, the month is almost over and I'll be closing it as promised. I had hoped someone else would chime in about any similar project they'd heard of, but nobody has. There have only been about four hundred thread views, so possibly this forum simply isn't visible enough. I will be querying elsewhere. |
ttk, why not fork centos6 yourself?
|
Quote:
|
Quote:
I tried forking Slackware in 2001. It was a lot of work. A lot of work. It proved more demanding than I could sustain, and RHEL(-cum-CentOS) is considerably more complex than Slackware. I'm simply not of Volkerding's caliber. I'm also interested in getting people of like minds together on it. If there are two or three independent groups out there all unaware of each other and all working on their own forks, I'd like to make them aware of each other for potential collaboration. Hence my query here. Frankly, the derision in this forum surprises me a little. Once upon a time, forking a distribution was considered flattering to the parent distribution. Having a lot of derivative distributions is a sign of vitality and appeal. The tone here suggests that is no longer the prevalent attitude. |
All times are GMT -5. The time now is 06:04 AM. |