Will systemd make FreeBSD more popular?

Status
Not open for further replies.
Many of my colleagues have decided just like me to stick with Red Hat (or clone of it) 6.XXX which will be available until 2020 and postpone the decision until dust settles.
Click to expand...

Technically, RHEL 6.x may be around until 2020, but is it really practically to stick with RHEL 6.x much longer? I think the kernel in RHEL 6.5 is already 5 years old.

Once Debian is on board with systemd, I expect Red Hat to accelerate its effort to make more software dependent on systemd.

Some people may stick with old Red Hat, or CentOS, or whatever. But I suspect most will feel it's time to move on, and just accept it.
 
Dies_Irae said:
Well, consider this thing from another point of view: we are moving towards an OS made of a single monolithic file. No config files, no libraries, and only one thing to keep up-to-date. Not to mention the install procedure... a dream for every sysadmin! :D
Click to expand...

I really wonder why systemd advocates don't just move to MS-Windows? If you want one big file that controls everything, controlled by a monopolistic corporation that is intent on controlling the entire ecosystem. If you don't mind vendor lock-in, or a subscription service, or having the inner workings of your OS hidden from you. And let's be honest, MS-Windows has the best desktop x86 hardware support in the business, and better graphics. If you are a PC end users at your place of work, Microsoft skills are far more widely used than everything else put together. Windows runs practically all widely used desktop applications, especially games.

After being a Linux advocate for decades, I now wonder if Linux has any place in the world? Maybe something like this makes more sense?
  1. FreeBSD for servers, or tech lovers.
  2. Windows for typical desktop/laptop end users.
  3. iOS or Android for hand-held devices
Linux is trying to fit into all three of the above categories, and I am not sure if it does an especially good job at any of them.
 
walterbyrd said:
I am not sure how far you can take that analogy.

Problem is, can you count on a lot of support for an OS that practically nobody uses?
Click to expand...
I forgot to reply to that myself but I feel the same way about the "nice, quiet beach". At times, I show my irritation on these boards for some who come over to investigate FreeBSD who have obvious Linux roots and take quick jabs like this was reddit or somesuch. The peacefulness and knowledge here has a lot going for it.

I have the same issue with a new client that I started a thread about last week. They are strictly a Windows shop but run a Linux server for their web site. These guys can barely spell "internet" much less know how anything works outside the Windows software they write for the devices they sell. I am going to tell them they need to switch web hosts because their current provider can't deliver the services I need as I update their web site along with other interfacing needs. One of their two arguments is "no one uses FreeBSD so who's going to support it?" My reply is, "Who do they call when their CentOS web server goes down?"

I'm going to have to update that thread cause an interesting thing happened last week. I'll try and do it today but our state governor and a 25-member economic development team is visiting today(!).
 
This might not be completely relevant, but I came across this entry which nicely explains some of the pains still to come with systemd. I was not aware of all of the things comming with your PID being 1, but there will be some interesting times for people who trust systemd.

So in the end it might make FreeBSD more popular, but the question really is if that is in our interest. I like quiet beaches, too.
 
In order to discuss about systemd, it would be neccesary to define what it IS. Not what it claims to be or aims to be, but what it is and what it most likely will become.
Then we can discuss about its merits.
 
Crivens said:
In order to discuss about systemd, it would be neccesary to define what it IS. Not what it claims to be or aims to be, but what it is and what it most likely will become.
Then we can discuss about its merits.
Click to expand...

Most importantly, not what people who are not going to use it or are even interested to find out what it is, claim it to be.
 
ANOKNUSA said:
There's no logic to this. This thinking is exactly why reasonable debate over systemd is virtually non-existent, and why this thread will inevitably be closed.
Click to expand...

Why would you say there is no logic to what I posted? I explained my reasoning. Systemd does make Linux much more Windows like. I have explained exactly why, and you have not refuted any of my reasons.

My post was not just a rant, I was not just spewing vitriol. As I explained: MS-Windows is superior to both Linux, and FreeBSD, in some respects. That is not an insult, it is a statement of fact. For example: nobody can beat Windows when it comes to new x86 desktop/laptop hardware compatibility.

Compared to Linux, Windows is more standardized, more monolithic, and more proprietary. Microsoft believes in protecting the end user from complexity by hiding the inner workings of the OS. Some people favor this approach - especially end users. Windows is very good with GUI control panels, and network configuration panels and the like.

For good or bad, systemd is more Windows like in all of those respects. And a lot of people are raving about systemd because of it.

Personally, I do not favor the MS/systemd approach. But I do wonder, if people like the systemd approach, why not just use Windows? Windows has all the advantages of systemd, and then some.

I think I can defend everything that I have asserted. I am not sure why you think my statements are so illogical and/or unreasonable; or why they would be cause to shut down the thread.
 
kpa said:
Most importantly, not what people who are not going to use it or are even interested to find out what it is, claim it to be.
Click to expand...

I am not certain that is entirely accurate. For example, what if new applications are written for systemd, and that makes the applications more difficult, or even entirely impractical, to port those apps to FreeBSD?

In that case: if you are a FreeBSD user, then systemd might affect you, even if you have no interest in ever using systemd.

I think we can get a pretty good idea about where systemd is heading by: 1) Considering Red Hat's motives, and: 2) Considering Red Hat's pattern of behavior so far.

Issue (1) is easy: Red Hat is a public corporation, and as such has a responsibility to it's shareholders to maximize profit. Clearly it would be advantageous to Red Hat if Red Hat controlled the Linux - and maybe even UNIX - ecosystem

Issue (2) might be more complex, and more debatable. Personally, I see Red Hat making a series of changes, even before systemd, that have no technical advantage, but those changes may make Linux more dependent on Red Hat standards. Systemd is still being advertised as an init replacement, by systemd if far more than that.

All JMHO, of course.
 
People running Ubuntu have at least until April 2019 before they face having to run systemd (Ubuntu is one of the few Linux distributions still holding out on systemd as the default init software). Debian probably has until at least 2017, perhaps longer if Wheezy gets LTS support like Squeeze did. Plus Debian may have the Devuan fork providing a non-systemd base for Debian packages.

As Oko said, there are options for people who do not like systemd besides FreeBSD. Of course, FreeBSD is an attractive choice, but it is one of several.
 
I usually stay out of the systemd conversation as I've admittedly little technical knowledge of the project and any conversation here(and elsewhere) usually revolves around "what ifs?" and personal opinion. I can only make the assumption more threads on this topic will pop up from time to time at least in the short term. Maybe it's just me, but what I would really like to see is someone/people that actually uses systemd in a production environment post some relevant information on what kinds of things it made easier for them or some of the technical pitfalls in using it has caused along the way. While I can't search the Internet constantly, I have yet to run into any generally objective posts in technical nature from that point of view. As far as how systemd will affect FreeBSD, I don't know but there are too many "what ifs" or lack of knowledge/experience with it at this point for me to form an informed opinion.
 
While it's a nice thought, I wonder how many of us working at FreeBSD-centric jobs would easily be able to find another one if the current one disappeared. Here in the Northeastern US, I see very few FreeBSD-centric jobs listed, the vast majority are Linux. I don't think that's going to change very quickly.
 
  • Thanks
Reactions: Oko
I have only an academic interest in systemd, as I currently do not have to use it and am also not dealing with applications which demand it. But it looms on the horizon that one day I may be forced to do it, and I like being forced to do things about as much as the next person. So I spend some time to see if it would be a good idea (nullifying the being-forced factor), but what I see there is nothing that convinces me as an engineer. Untill now, I have only seen that systemd tries to solve things that are already solved, and the extra benefit it brings to the table do not justify the risk of touching that part of the infrastructure. That is my opinion.

Yesterday I had a small conversation with one systemd defender, and he said that he liked the service files and the journaling in binary. But he could not express to me why a binary log would be superior. It logs a lot more meta data, yes, but here is the real-world admin yet to speak up on the positive points of that. Does anyone know such an administrator to ask for input?

Some other facts I can dig up using a stop watch: my systems use between 2 and 4 seconds from loading init untill they reach the login prompt. That time might be cut in half by a smarter init system, but these are systems tuned to be of personal use. How long does a real server take to reboot, and what could be saved by making the startup faster? I do not have such machines, and if I had I might be uninterested in rebooting them only to watch the console holding a stop watch. Maybe someone here can share some numbers on that, but for me the time saved is in no relation to the time it would take me to set up some other init system. For each minute spent there, I can reboot my system for a month, each day, untill I get to the break even point. And this posting would equal a year of reboot time saved.
 
How long does a real server take to reboot, and what could be saved by making the startup faster?
Click to expand...

I have used CentOS 6.5 (no systemd) and CentOS 7 (which has systemd) on the same box. I was surprised to find that the CentOS 7 took more time to boot. A systemd advocate told me that was because I was not running many services. Systemd's great strength is that it starts many services in parallel.

Bottom line: the time you may, or may not, save depends on your particular setup. I can see where it would be easy to rig benchmarks, either one way, or the other. Also, there are other methods of increasing boot-up speed without the massive system-wide changes of systemd. How those other methods compare to systemd is probably a matter of the individual systems. Something else that may need to be figured into the equation: a systemd setup may require more booting than standard init setup (just something I have read).

Personally, it is difficult for me to believe that boot-up time would be such a huge issue that massive system-wide changes would be justified. A well maintained Linux service may only need to be rebooted once every few months. Also, I am not convinced of the safety in starting a lot of services in parallel.
 
I have compared Debian's upcomign version 8 (no systemd) with Debian 8 (with systemd). There was no difference in boot time, a few services/daemons didn't work with sysetmd init. Unit (service) configuration was a little easier to parse, but most administrators do not need to adjust init scripts anyway. Binary logging was really inconvenient, but it can be turned off. For desktop users there probably isn't any practical difference. For system administrators systemd is more trouble than benefit, in my opinion, but it isn't a huge issue.
 
NewGuy said:
People running Ubuntu have at least until April 2019 before they face having to run systemd (Ubuntu is one of the few Linux distributions still holding out on systemd as the default init software).
Click to expand...

Ubuntu is based on Debian, and therefore follows Debian. Debian will default to systemd, probably, a little later on this year.

Furthermore, when you say you have until 2019 to switch. Remember: your system may, technically, be supported until 2019; but that does not mean you will have an up-to-date system that will run recently released applications. Supposedly, CentOS 6.5 will be supported until 2020, but the kernel is already five years old, and I was have trouble with some recent applications just a few months ago.

Ubuntu's great appeal is to desktop users. Desktop users usually want the latest versions of applications. I think this is why Ubuntu does not use Debian stable. For most desktop end-users, the latest version is more important than being super-stable. I think it is doubtful that many Ubuntu users will want a five year old OS in 2019, or that many CentOS users will want a ten year old kernel in 2020. Red Hat has made it clear that they intend to have systemd take over even more. Want the newest version of udev? You need systemd. Might not be a problem now, but what about two years from now?
 
For system administrators systemd is more trouble than benefit, in my opinion, but it isn't a huge issue.
Click to expand...

Not a huge issue? Are you kidding? Systemd is a radical departure from traditional Linux, and promises to change Linux much more radically in the near future. Systemd is still in it's infancy. Once Debian is firmly established with systemd - about a year a from now - Red Hat will control Linux like Microsoft controls Windows.

With systemd, POSIX, and the UNIX philosophy are gone.

With systemd, you can forget much of what you know about CLI, configuration files, logs, and much more.

No more: init, login, PAM, getty, syslong, udev, mount, cryptsetup, cron, at, dbus, acpi, cgroups, gnome-session, autofs, tcpwrapper, audit. No more run levels, no more text logging. And systemd is just getting started.
 
I've been using FreeBSD for 2-3 years for various things, and ramped up my use after Arch broke all non-systemd installs by effectively making everything dependent on libsystemd.

I don't want FreeBSD to become more popular and take the role that Linux has been serving - that's not what I'm about. FreeBSD has just about enough support for me to be satisfied with its use.

As far as Jordan Hubbard is concerned, he can shove launchd where the sun doesn't shine - I'm not an Apple user, and I don't want any more of their franken-Mach bull in my OS. OS X is more Mach than BSD, and its not something I aspire to use - I left OS X because Apple is nothing more than a status symbol - OS X is just as bad as Windows, if not worse.

If FreeBSD gets launchd, by that time my NDAs will be done and over with, I'll personally fork FreeBSD and invite anyone to join me. I already have voiced before that OpenRC and Runit are both fantastic options for a new init system, and that we're a BSD, not an Apple OS.
 
walterbyrd said:
Ubuntu is based on Debian, and therefore follows Debian. Debian will default to systemd, probably, a little later on this year.

Furthermore, when you say you have until 2019 to switch. Remember: your system may, technically, be supported until 2019; but that does not mean you will have an up-to-date system that will run recently released applications. Supposedly, CentOS 6.5 will be supported until 2020, but the kernel is already five years old, and I was have trouble with some recent applications just a few months ago.

Ubuntu's great appeal is to desktop users.
Click to expand...

First, yes, Ubuntu is based on Debian, but the first stable release of Ubuntu to be based on a version of Debian with systemd will not happen until April 2016. So that gives people over a year. And, again, supported releases with Upstart are supported through to 2019.

As for up to date software, Ubuntu and Debian have backports, meaning users on older system will have access to modern applications during that time. As for kernels, you only need a newer kernel if you are running newer hardware that isn't supported. Existing hardware will continue to work and it is easy enough to purchase only hardware supported by existing kernels.

Further, while Ubuntu has a strong following on the desktop, most desktop users won't be affected by systemd. I was talking about systemd's impact on servers (where Ubuntu is also quite popular these days), and people running servers rarely want to stick on the cutting edge.

Which brings me back to my previous point, people in the Linux community (especially those running CentOS, Debian and Ubuntu) have years to figure out what they want to do with regards to systemd.
 
ANOKNUSA said:
There's no logic to this. This thinking is exactly why reasonable debate over systemd is virtually non-existent, and why this thread will inevitably be closed.
Click to expand...
Every thread seems to go this way, I think. I was ready to write a response to some of the ridiculous statements that have popped up, but why bother? There is some good information in this thread to go with some nonsense, at least. I just wish people would listen and try to actually understand what Jordan said in his presentation, but oh well. I'm going to go ahead and at least post it again:
 
Okay, I did an install of the currently available Ubuntu on a VirtualBox, since I need that one for my job. Some projects have so deep links to Linux that they do not build outside of that ecosystem, and I have not the time to patch that up. But I was checking for systemd and yes, it is running. On a current Ubuntu. So now I can check some things I want to see on that virtual machine.

Lets discuss things further, to see if we can get to the point where we can find answers and maybe have something like a plan, before the thread diverges and gets closed. That would be a bit useless in my humble opinion, as a new one will instantly respawn and continue. We need more data to support a decision. My gut tells me what I would like to see, but my engineering mind wants these pesky details... and I hope I am not alone with that.
 
Crivens said:
Okay, I did an install of the currently available Ubuntu on a VirtualBox, since I need that one for my job. Some projects have so deep links to Linux that they do not build outside of that ecosystem, and I have not the time to patch that up. But I was checking for systemd and yes, it is running. On a current Ubuntu. So now I can check some things I want to see on that virtual machine.
Click to expand...

Ubuntu does not yet use systemd in its stable releases. There is a systemd shim, but the init software in Ubuntu's stable branch is still Upstart, specifically Upstart version 1.12.1. Development versions of Ubuntu may have systemd.
 
TeamBlackFox said:
I've been using FreeBSD for 2-3 years for various things, and ramped up my use after Arch broke all non-systemd installs by effectively making everything dependent on libsystemd.

I don't want FreeBSD to become more popular and take the role that Linux has been serving - that's not what I'm about. FreeBSD has just about enough support for me to be satisfied with its use.

As far as Jordan Hubbard is concerned, he can shove launchd where the sun doesn't shine - I'm not an Apple user, and I don't want any more of their franken-Mach bull in my OS. OS X is more Mach than BSD, and its not something I aspire to use - I left OS X because Apple is nothing more than a status symbol - OS X is just as bad as Windows, if not worse.

If FreeBSD gets launchd, by that time my NDAs will be done and over with, I'll personally fork FreeBSD and invite anyone to join me. I already have voiced before that OpenRC and Runit are both fantastic options for a new init system, and that we're a BSD, not an Apple OS.
Click to expand...

It's attitudes like this that will keep FreeBSD way behind the puck as markets fundamentally change. If you want more people to use our code or to even stay relevant, we to have to learn to adapt to current trends. I'm all for adopting something like launchd as long as it doesn't change FreeBSDs overall structure and design ethos. Think mobile, for example. Something like launchd could greatly improve projects like PC-BSD where targeting the mobile space (tablets/laptops) are more of a challenge. If you've watched Jordans talk, he states more Unix machines are running on batteries than any other form of power, which is true.

Just my two cents.
 
Beastie7 said:
it's attitudes like this that will keep FreeBSD way behind the puck as markets fundamentally change.
Click to expand...
You've just stated that FreeBSD is way behind the puck and nothing could be further from the truth.

This is not a popularity contest and any company that determines which OS they use based on a forum poster's comment is a company that won't be around a year from now. Popularity does not determine technical excellence or success. Nor do changes in the market necessarily make an operating system unable to handle its needs.

You can put FreeBSD up against any other operating system out there and be wildly successful. Ask one of its biggest users and contributors Netflix why they switched to FreeBSD just a couple of years ago.
 
Status
Not open for further replies.
Back
Top