How is FreeBSD coping with a systemd future?

Status
Not open for further replies.
Oko said:
Which technology you think I am going to trust :q

That might work for you very well. But age is hardly the only, or a factor at all for enterprises currently using the Linux ecosystem and seeing FreeBSD as irrelevant because there is nothing their current setups cannot do that FreeBSD can, and time & money required for the switch are absolutely not paying off eventual benefits gained. There's another factor that is, unfortunately, winning, and that's the fact that there are orders of magnitude more people using, fixing and coding new features for the Linux virtualization subsystems, Docker included. There will hardly be any mass exodus over systemd, between Linux distros let alone toward the BSDs. The group against systemd in the Linux ecosystem is very vocal, but unfortunately in significant minority.
 
I would not be so certain... I can imagine that paragraph posted in a Linux forum resulting in several users and maybe even some datacenter managers proceeding to google "this vs that" -- and as a result moving their machines over to FreeBSD... the only thing stopping that now being the idea of an alternative, as in "FreeBSD? it has a forum even? "
 
jb_fvwm2 said:
I can imagine that paragraph posted in a Linux forum resulting in several users and maybe even some datacenter managers proceeding to google "this vs that" -- and as a result moving their machines over to FreeBSD... the only thing stopping that now being the idea of an alternative, as in "FreeBSD? it has a forum even? "

I'm assuming you're referring to my paragraph. But perhaps you are not familiar how enterprises work. I'm pretty sure nobody is going to "move over their machines to FreeBSD" just because they read something in a forum, or blog post, unless their business does not depend on a stable and well understood environment. They might "fire up FreeBSD alongside their existing ecosystem" to see what is it all about. That's what we're doing.

We are a small company, delivering some SaaS solutions and managed hosting, few dozen servers, nothing big. We are running FreeBSD for a year now alongside existing Debian, Ubuntu, CentOS and Gentoo installations because we're testing and trying it out. We are certainly not going to base any decision without extensive testing and strategic planning. It's not just installing and using the system. We have tools, configuration templates and systems, monitoring, failover and redundancy, all tailored for the systems we use, all requiring change of some kind in order to "switch". I can imagine much bigger companies having even more resistance, testing requirements and above all paperwork, to change.

Above all, our testing for our use cases is telling us we can't switch.

Linux is faster for us because of PostgreSQL regression on FreeBSD 10. We're a PostgreSQL shop. Linux is more secure for us because we run well documented and understood SELinux and AppArmor, and the Linux kernel has had for years some security features (like ASLR) that are yet to come in FreeBSD (10.1), and MAC/RBAC on FreeBSD is completely unknown to us and pretty much undocumented. Linux as a distro with supported non-base software is more stable for us because it is stabilized by many, many people before release and there is much less work for us to test and deploy updates as opposed to FreeBSD ports which are moving as fast as a rolling Linux distro would, which requires far more testing, more manpower, build servers, staging/testing servers, etc... We're a Python shop and our software is grown on Debian 7 with Python 2.7.3, while on FreeBSD the 2.x branch is 2.7.8. Now, believe it or not, we can't switch without changing the code because some bugs were fixed and some features introduced between those two versions that is breaking our code on FreeBSD. That, of course, is absolutely not FreeBSD's fault, I'm just trying to show that there are many unforeseen problems and "switching over" is not just a matter of firing up a new VPS... Which in itself would also be a problem because our hosting company is not supporting FreeBSD. And if you've been in the web hosting industry as a consumer of hosting services as long as we have, you'd know that once you settle with a hosting company that works for you, finding another (that might support FreeBSD) is a nightmare that also requires many months of testing under various circumstances to see how they fare.

We can use Debian 7, Ubuntu 14.04 LTS, CentOS 6.x for years to come without having to violate our systems with systemd, and I'm pretty sure our clients would like us NOT to change anything in that department without significant amount of objective reason. We will continue to use FreeBSD in parallel and test it, perhaps there are use cases and scenarios which we haven't yet encountered that would significantly swing the opinion over for the switch.

Don't get the above as something against FreeBSD. Quite the contrary, despite all that, we continue to use FreeBSD in testing. I'm just trying to paint the picture how "switching over to FreeBSD" is far easier said than done. I'm sure the bigger the enterprise, the bigger the inertia toward any kind of change. So no, I don't see any mass exodus happening over systemd to anything - another Linux or BSDs.
 
AzaShog said:
I'm assuming you're referring to my paragraph. But perhaps you are not familiar how enterprises work. I'm pretty sure nobody is going to "move over their machines to FreeBSD" just because they read something in a forum, or blog post, unless their business does not depend on a stable and well understood environment. They might "fire up FreeBSD alongside their existing ecosystem" to see what is it all about. That's what we're doing.
I concur!

AzaShog said:
We are a small company, delivering some SaaS solutions and managed hosting, few dozen servers, nothing big. We are running FreeBSD for a year now alongside existing Debian, Ubuntu, CentOS and Gentoo installations because we're testing and trying it out. We are certainly not going to base any decision without extensive testing and strategic planning. It's not just installing and using the system. We have tools, configuration templates and systems, monitoring, failover and redundancy, all tailored for the systems we use, all requiring change of some kind in order to "switch". I can imagine much bigger companies having even more resistance, testing requirements and above all paperwork, to change.
The same here. A small academic lab couple of dozen physical servers, two dozen virtual, and abut 2-3 dozen of desktop machines. It took me a year to rebuilt it after taking over the infrastructure which was falling apart. Even though I had complete freedom to do anything I want it was difficult working around whatever existed because people had to continue doing their work. They were also got used to some things and taking those things away was as difficult as taking a candy from a my own kids.

AzaShog said:
Above all, our testing for our use cases is telling us we can't switch.
+1 Being an ardent OpenBSD users it was easy to switch entire existing network infrastructure (Firewalls, DNSs, LDAPs, monitoring, shell gateways, and few other things to OpenBSD). Linux has nothing worth looking in that domain. Then I start thinking about switching storage space to either FreeBSD/FreeNAS or DragonFlyBSD. I did lots of testing in-spite of the fact that DragonFlyBSD is very dear to my heart and I used to work with FreeBSD a lot in the past. One of the hardest thing to admit was that DragonFlyBSD is not usable in production even in a small shop. One great (arguably the best) file system is not enough reason to compensate for the lack of basic infrastructure elements (LDAP client not working), lack of basic monitoring tools and a friendly, knowledgable but tiny community which lacks critical mass. FreeNAS and FreeBSD on another hand prove to be worthy adversary to Linux as a data storage OS and I switch all but one well configured file servers from Red Hat to FreeNAS. There was just no incentive to spend 2-3 days messing with it as Red Hat was working well enough on the hardware RAID. Then I start looking at the virtual hosts and due to the fact that our primary guest OS is Linux (due to heavy use of Java in our proprietary product) two days of testing VBox and half a day of reading about the state of Bhyve was enough to realize that we are KVM do or die shop. Computing nodes, desktops and similar was an easy decision. I need MATLAB to work. MathWorks doesn't support MATLAB for FreeBSD. That was it.

AzaShog said:
Linux is faster for us because of PostgreSQL regression on FreeBSD 10. We're a PostgreSQL shop. Linux is more secure for us because we run well documented and understood SELinux and AppArmor, and the Linux kernel has had for years some security features (like ASLR) that are yet to come in FreeBSD (10.1), and MAC/RBAC on FreeBSD is completely unknown to us and pretty much undocumented.
Coming with the OpenBSD background I have a very low opinion of MAC as a general idea and even lower opinion as a practical idea but I will take your word it. On my key network infrastructure points I am happy to settle with the OS which has a code base smaller of order 20-30 times than Linux. Somehow makes me feel there are fewer bugs in the 30 times smaller code base ;)

AzaShog said:
Linux as a distro with supported non-base software is more stable for us because it is stabilized by many, many people before release and there is much less work for us to test and deploy updates as opposed to FreeBSD ports which are moving as fast as a rolling Linux distro would, which requires far more testing, more manpower, build servers, staging/testing servers, etc...
You are not serious about this one §e ? Red Hat 7.0 formally came in June. There are still no EPEL RPMs released for it. QA is pitta comparing to OpenBSD. While I concur that FreeBSD 10.0 feels little bit like a ride on the wild side at least it has 24000 ports comparing to 0 RPMs for Red Hat 7.0. 9.3 is definitely very stable just like the Red Hat 6.5 which we use.

AzaShog said:
We're a Python shop and our software is grown on Debian 7 with Python 2.7.3, while on FreeBSD the 2.x branch is 2.7.8. Now, believe it or not, we can't switch without changing the code because some bugs were fixed and some features introduced between those two versions that is breaking our code on FreeBSD. That, of course, is absolutely not FreeBSD's fault, I'm just trying to show that there are many unforeseen problems and "switching over" is not just a matter of firing up a new VPS... Which in itself would also be a problem because our hosting company is not supporting FreeBSD.
We are Python shop as well. I can't believe what you just said. If you proprietary code brakes from 2.7.3 to 2.7.8 you have much bigger problem than picking an OS for your shop. Red Hat 7.0 ships with 2.7.5 and I am sure by the first "usable" 7.1 or 7.2 release it will updated to 2.7.8. My main grudge with Red Hat 7.0 is systemd. It is not so much that breaks my services and chkconfig routine (finally I am relative new comer to Linux using True64, Solaris and BSDs circa 1990)
but the crappy systemd code and 0 absolute 0 quality assurance for such a monumental and profound change how the system works is frightening. We will certainly not switch to 7 branch until the end of life of Red Hat 6 branch and if FreeBSD gets Java (somebody told me that foundation signed the contract with Oracle and possibly MATLAB) we will most definitely switch to FreeBSD. I would really, really like to see HAMMER ported to FreeBSD too. I still hope that Red Hat gets to its senses and just remove systemd.

AzaShog said:
And if you've been in the web hosting industry as a consumer of hosting services as long as we have, you'd know that once you settle with a hosting company that works for you, finding another (that might support FreeBSD) is a nightmare that also requires many months of testing under various circumstances to see how they fare.

We can use Debian 7, Ubuntu 14.04 LTS, CentOS 6.x for years to come without having to violate our systems with systemd, and I'm pretty sure our clients would like us NOT to change anything in that department without significant amount of objective reason. We will continue to use FreeBSD in parallel and test it, perhaps there are use cases and scenarios which we haven't yet encountered that would significantly swing the opinion over for the switch.
Couple hard core UNIX users could definitely sway opinion in your company. I find an average Linux system admin completely ignorant of UNIXes (Solaris, AIX, HP, BSDs). Customers are another story and I find them very prone to PR propaganda. I put monumental fight to keep away Ubuntu and Debian away from our shop and typical line of attack against Red Hat was always "it has older kernel" and "packages are outdated". Nothing of course can be further from the truth as Red Hat heavily back ports their kernels and people who know little bit about RPMs run the same versions of software on RedHat as on Ubuntu or Debian.

AzaShog said:
Don't get the above as something against FreeBSD. Quite the contrary, despite all that, we continue to use FreeBSD in testing. I'm just trying to paint the picture how "switching over to FreeBSD" is far easier said than done. I'm sure the bigger the enterprise, the bigger the inertia toward any kind of change. So no, I don't see any mass exodus happening over systemd to anything - another Linux or BSDs.

I wish I could say the same for DragonFlyBSD. I will wait to see what comes out of HAMMER2 but speaking of HAMMER my only hope for production use is that it gets ported over to FreeBSD. That dma and Sasha's work on kernels in the userland will be nice too :)
 
Oko said:
You are not serious about this one §e ? Red Hat 7.0 formally came in June.

Having run several distros, and now with FreeBSD, in parallel for years, yes, I'm dead serious. RHEL 7 might have been just released, but we don't use RHEL, and CentOS 6 that we do (because clients require it) is several years old now since its pre-release feature freeze. The same applies to Debian 7 and Ubuntu 14.04 LTS. All those systems are feature-frozen and well tested by many, many people before we run our own tests. And this includes non-base software. What those systems guarantee between releases is that major version changes in software will not happen. That change in the configuration files, if any at all, will be backwards compatible and not required.

Just an example of how this is important and how this bit us on Gentoo (meaning FreeBSD's ports are sensitive to this issuse too). A company we managed the systems for was running some software based on ImageMagick. Some image processes broke because Gentoo upgraded ImageMagick, which changed default blurring parameter constants in the new (major) version, which wasn't caught by automated tests because no software or API was broken. It was caught many weeks later. That would never have happened on a non-rolling release distro because you test the new release from all the angles for weeks/months anyway. It just physically can't be done on a rolling release distro.

I can't believe what you just said. If you proprietary code brakes from 2.7.3 to 2.7.8 you have much bigger problem than picking an OS for your shop.

I never said we didn't. It's a known bug that is going to be fixed. I was just giving an example how totally unexpected situations may arise and switching over to another OS is far from trivial. It takes months of testing, rewriting your standard procedures, utility tools and above all re-training or hiring new manpower for something like switching from a Linux to a BSD.
 
Is systemd still an issue with FreeBSD? It is a hot potato topic on forums like OpenSUSE, or Ubuntu, because both distributions have drunk the Kool-Aid.
 
Last edited by a moderator:
Is systemd still an issue with FreeBSD? It is a hot potato topic on forums like openSUSE, or Ubuntu, because both distributions have drunk the kool-aid.

It's only relevant if it breaks upstream ports due to dependencies, and there's an increasing groundswell working to mitigate that. I think the wheels will fall off the systemd juggernaut soon. It's expanding far too aggressively, with little scope or specification, and it will surely start to fail under its own complexity.
 
Agreed, systemd is likely to have very little impact on FreeBSD, especially server instances. I could see it affecting a handful of ports desktop users like (especially software related to GNOME), but I don't think it is going to be a big issue. For most things we can port around systemd or put compatibility libraries in place.

By the way, I entirely agree with AzaShog. FreeBSD is a great operating system, but it is a big undertaking moving from one operating system (like CentOS, Debian or Ubuntu) to another. I work in a small tech shop and spent months testing and running trials. Eventually we found we could switch to FreeBSD 10.0, but it would require a number of changes and setting up FreeBSD with the services we needed took a lot longer than setting up the same environment with Debian/Ubuntu. And then FreeBSD 10.1 came out and stopped working on one of our servers, so upgrading was a dead end. And we're a small shop, fairly flexible. An enterprise level company is going to stick with the idea of "it's not broke, don't fix it".
 
Last edited by a moderator:
An enterprise level company is going to stick with the idea of "it's not broke, don't fix it".
And that's a good policy for production systems. I'm working in a shop using Microsoft .NET for everything and that's ok, it's a decent framework allowing for a lot of rapid development.

For my private boxes though, I always used Debian GNU/Linux, as it's free (although, unfortunately, NOT like in "beer" *g*), manageable (even in spare time) and does everything I need. In fact, systemd was THE issue bringing me here¹ ... having my share of pain with pulseaudio and seeing another big moloch by the same author on its quest to conquer and assimilate systems -- I thought it was about time to try something different. FreeBSD came to mind immediately. My very first project is getting my notebook operated using FreeBSD, as it is not a critical system. I'm struggling with wifi connection at the moment, I already bought a mini-pcie card that should be supported, but I'm facing kernel panics. Not giving up right now, though. My notebook is undergoing a make buildworld from svn right now....

Me, too, I don't think systemd will become a huge problem for FreeBSD. Although it looks like GNOME doesn't care much about portability any more, KDE certainly does, and hopefully a lot of other desktop software. Hard systemd dependencies are probably only an issue with desktops and maybe they will come down to only systemd-logind, which means that in the long run, a service on FreeBSD could be needed implementing the dbus interface of systemd-logind for some of the desktop-related ports. Of course I can't know for sure without looking at this interface, but my impression ist: Can't be TOO hard.

¹ That could be interpreted as a (strange) way of systemd actually helping FreeBSD and other *BSDs (by increasing their user base), at least when I'm not the only one looking for Linux alternatives because of systemd.
 
That could be interpreted as a (strange) way of systemd actually helping FreeBSD and other *BSDs (by increasing their user base), at least when I'm not the only one looking for Linux alternatives because of systemd.

For quite a number of months I have posted on Reddit that I noticed this where I would get downvoted into oblivion and made fun of that this isn't happening at all. (Not that I value the opinions or votes of Reddit. It's one of two times I ever posted there.)
 
What strikes me about those systemd topics (and this board in general) is the amount of attention the BSD community gives to Linux. What's going on here - Stockholm Syndrome maybe. At this point systemd is a mess but end of life of RHEL/CentOS 6 and SLES 11 is somewhere in mid 2020s (other distros are as irrelevant as BSD in the enterprise area). So no - there won't be any massive movement toward BSD, at least not in the next few years (I'm not talking about basement deployments of a LAMP stack here). And then, I've never heard about anyone switching from AIX just because he did not like ODM.
 
What strikes me about those systemd topics (and this board in general) is the amount of attention the BSD community gives to Linux. What's going on here - Stockholm Syndrome maybe.
At this point systemd is a mess but end of life of RHEL/CentOS 6 and SLES 11 is somewhere in mid 2020s (other distros are as irrelevant as BSD in enterprise area). So no - there won't be any massive movement towards BSD, at least not in the next few years (I'm not talking about basement deployments of a LAMP stack here).
And then, I've never heard about anyone switching from AIX just because he did not like ODM.
1. EoL for RHEL and SLES may be in 2020s, but not so for Debian 7, which is the last with SysVinit.
2. FreeBSD developers want systemd-like init too: http://www.phoronix.com/scan.php?page=news_item&px=MTg0ODE
 
"We need to be open to fundamentally new approaches and ruthlessly cull what is no longer demonstrably useful to the 99%"
 
"We need to be open to fundamentally new approaches and ruthlessly cull what is no longer demonstrably useful to the 99%"
zspider, when I put your posting next to the Blockupy movement, interesting thoughts begin to take shape ;)

But to add to the topic - the attention brought towards Linux is mainly (in my humble opinion) based in the fact that most of our application upstream is there. Should something go on there that would cut off any chance to use new versions of (for example) KDE, any KDE user in *BSD would be in trouble. It is only natural for humans to keep an eye on possible threats, real or imagined. Also, Linux is kind of a tidal pool where some evolution is going on and which is being reset at a regular basis. It might be interesting to watch - but not interesting to enter. Wow, that analogy really seems to fit :)
 
What strikes me about those systemd topics (and this board in general) is the amount of attention the BSD community gives to Linux.
Maybe because many of us run Linux for living and can't believe one after another non-sense coming out of RHEL kitchen.

So no - there won't be any massive movement toward BSD, at least not in the next few years
I would not be so sure. It took Linux only 15 years and tens of billion of dollars investment from Silicon Graphics, Compaq, HP, IBM, Red Hat, Novell, and Canonical to come close to Solaris. Current direct investment in BSDs is measured and hundreds of thousands. Let see one/two large corporations standing seriously behind one of BSD projects.
 
However the entire reason I started this thread is that until that OpenBSD GSOC project nobody was moving forward on said workarounds and Desktop Developers are already moving forward on shifting away from Consolekit (As far as I know Gnome is now shipping in degraded mode for systems not supporting the logind dbus interface) and distribution specific code towards systemd's interfaces.
This is because FreeBSD developers unlike OpenBSD developers are running OS X on their desktops so as far as they are concerned anything desktop-related is irrelevant.
 
zspider, when I put your posting next to the blockupy movement, interesting thoughts begin to take shape ;)

But to add to the topic - the attention brought towards Linux is mainly (in my humble opinion) based in the fact that most of our application upstream is there. Should something go on there that would cut off any chance to use new versions of (for example) KDE, any KDE user in *BSD would be in trouble. It is only natural for humans to keep an eye on possible threats, real or imagined. Also, Linux is kind of a tital pool where some evolution is going on and which is being reset at a regular basis. It might be interesting to watch - but not interesting to enter. Wow, that analogy really seems to fit :)

Actually that was a direct quote from the slideshow on the Phoronix article. :)
 
But to add to the topic - the attention brought towards Linux is mainly (in my humble opinion) based in the fact that most of our application upstream is there.
I think this is exactly the point. Supporting an operating system, you have to take into account that it's the applications getting the work done, the OS is just the infrastructure for them. As such, it is very important, but doesn't do anything useful by itself. So if some of these applications start depending on things only available on one specific OS, this is where the trouble starts...

FreeBSD developers want systemd-like init too:
Hmm, there's probably nothing wrong with modernizing init. Existing inits have their flaws (e.g., on my FreeBSD installation, service hald stop always leaves some hald processes running). But implementing new interfaces, pulling in lots of responsibilities through modules using these interfaces, and having it all tightly coupled to a specific OS; I think this is clearly NOT the way to go.
 
This is because FreeBSD developers unlike OpenBSD developers are running OS X on their desktops so as far as they are concern anything desktop related is irrelevant.

While maybe true for some developers, this is obviously not the case across the board or there wouldn't be a use for a Desktop Usage section here on the Forums. ;)
 
And GNU userland isn't Unix either.

That's my point. Why should FreeBSD be concerned about the Linux proprietary systemd when large swaths of themselves don't use it?
Code:
Proprietary software or closed source software is computer software licensed under 
exclusive legal right of the copyright holder with the intent that the licensee is given the right 
to use the software only under certain conditions, and restricted from other uses, such as 
modification, sharing, studying, redistribution, or reverse engineering. Usually the source code 
of proprietary software is not made available.
Screenshot from 2014-11-27 14:28:54.png
 
What strikes me about those systemd topics (and this board in general) is the amount of attention the BSD community gives to Linux.
Without adding up the numbers, most of these threads come from Linux posters on this board, or posters with post counts of one or two or so which makes me question why they posted here. Otherwise, threads about Linux are fairly rare here so, no, this BSD community does not give much attention to Linux.

As far as I care, they could all be deleted because they are usually pointless to the BSD community.
 
Without adding up the numbers, most of these threads come from Linux posters on this board, or posters with post counts of one or two or so which makes me question why they posted here. Otherwise, threads about Linux are fairly rare here so, no, this BSD community does not give much attention to Linux.

As far as I care, they could all be deleted because they are usually pointless to the BSD community.
That is also my impression.

"Fear is the path to the dark side. Fear leads to anger. Anger leads to hate. Hate leads to trolling BSD forums to spread FUD".
 
Like it or not, systemd has already affected FreeBSD. There are more than a few new FreeBSD users because of it, for one thing. It has also forced a reevaluation of init systems, and brought up the question of how FreeBSD can be compatible with software that is written to depend on a Linux monoculture.
 
Status
Not open for further replies.
Back
Top