Docker is dead

THANK YOU!!!

Finally a pragmatist ignoring the NIH bias

Well you seem to have missed the point, most likely intentionally. It has nothing to do with not being invented by FreeBSD. That's an idiotic comment. Do you know FreeBSD has been using GNU userland tools up until only recently, or provides Linux emulation? No you conveniently disregard the facts in your head-first evangelical salesmanship of Linux containers.

bakul may be right that it will attract more corporate users to FreeBSD, but so what? Prove it.
Unlike Linux, and as I wrote earlier, FreeBSD does not have the man-power and funding to run off emulating everything that comes out ofthe huge "corporation" that is Linux.


datacenter resources to be able to grow rapidly and predictably on very stable infrastructure. And once you have that assurance and capability, there are very few reasons to go back to in-house big iron management, especially when you know it's a never-ending battle of upgrading hardware and middleware in a brittle in-house datacenter that will NEVER have the redundancy and safety of dedicated providers
You really do think the world revolves around cloud computing? I've never read so much guff; you've got to be a cloud salesman surely?
This space is already taken by Linux. Why waste developer's time pursuing this to just be another, lesser known provider? Where's the rate of return in this? You might be a good salesman but you stink at accounting... :)

Don't get me wrong, I admire your tenacity, but steadfast ignorance of reality isn't doing you any favours. :rolleyes:

I wrote earlier and I will write again, put your efforts into convincing the Linux foundation to fund it.💰
 
Is NIH an initialism? What does NIH stand for? I've never heard of it.
Not Invented Here [syndrome]

It is often derogative to suggest that someone just wants to hack on their own code rather than integrate with existing software. However in practice it can also be very useful to be in the position to reimplement the solution entirely if it is a mess.

The usage can be:
I suffered a bit from NIH because that string splitting function we were originally using was dragging in 12 different scripting languages and 100+ libraries. Now it uses only 5 lines of code and zero deps.
 
  • Thanks
Reactions: a6h
bakul may be right that it will attract more corporate users to FreeBSD, but so what? Prove it.
prove would be some of my clients then, though that number is negligable. Probably even more companies at least could change their reproducible CI/CD toolchain and also their server fleet to FreeBSD then => more users = more feedback/quality improvement and finally more software vendors stop ignoring FreeBSD. Companies starting over would at least have a choice and not be forced to use Linux. It might not be what the majority of this discussion want, but there are some users like me who do.

I wrote earlier and I will write again, put you efforts into convincing the Linux foundation to fund it.💰
Why should anyone try to convince the Linux Foundation to fund a (probably non-GPL) implementation on FreeBSD? I would not even ask a fancy glittering startup like Docker Inc. for this (the implementation of docker sucks, they should have done it the podman way right from the start), but the FreeBSD Foundation. The concept of application containers is nothing Linux specific but just something jails lack - and before the whole discussion starts all over again about why application containers are different than jails I mean the whole infrastructure to work with those containers namely declarative container configuration, private hub/repo, syncing deltas etc. (and for those who would like to suggest scripting the whole thing and provide jail tarballs on nginx: NO, just NO)
 
To piggyback. There is no containerization in Docker. It's cgroups and namespaces, that's it. Security is not even a design constraint for docker.

Docker should be called CFEngine++.
 
Either way, I'd be interested in seeing, how do you work around the main part that breaks most docker images on FreeBSD; ie the library/executable incompatibility... As either way we go, putting a wrapper over Jails are not the problem, but more of the fundamental issue that linux can't use FreeBSD pre-compiled stuff, and there is only partial support the other way around. So you are stuck having the entire docker image(all the way from and including the base image of Kali or other base linux distro) being recompiled every time you want to you run; which means you can't just "spin up" another docker image and have it working in minutes.
 
What we should really do is not care what Linux does. What does it matter? We do our thing on FreeBSD and do it better than they can. So...don't care.

That should be a topic category for things like this. Don't Care.
 
Either way, I'd be interested in seeing, how do you work around the main part that breaks most docker images on FreeBSD; ie the library/executable incompatibility...
You are basically asking why you can't run a program compiled for one operating system on another operating system.

People have been asking this question since MS-DOS days (and earlier). The answer is: You can't. Not properly. It is neither operating systems fault. This is life.

Docker has overreached slightly and said it can. But ultimately it relies on emulating one operating system. Linux. One operating system, these forums are specifically not about ;)
 
I'm well aware we can't run stuff from one OS on another; but in the end; the stuff running on Docker is compiled to run on Linux. If anyone wants to get docker to run on FreeBSD, they will have to solve the fundamental issue (which like you said, can't be done properly). The only way, is giving up all the advantages that Docker gives. Even if, you do the ugly solution of putting if statements in the docker file, and detect which OS you are running, doesn't work in the end. As you are still back to be beginning, of not having a common base that everyone has.
 
Is NIH an initialism? What does NIH stand for? I've never heard of it.
Not Invented Here

edit: I just see my browser didn't show me a bunch of newer replies :eek: But, adding to kpedersen's reply, NIH also applies to standards and concepts, not only to existing software ;) (that's what I was referring to when talking about Powershell in this Linux thread. MS doing their own implementation probably made sense, but they could have followed the POSIX shell standard as a common baseline)
 
  • Thanks
Reactions: a6h
Either way, I'd be interested in seeing, how do you work around the main part that breaks most docker images on FreeBSD; ie the library/executable incompatibility... As either way we go, putting a wrapper over Jails are not the problem, but more of the fundamental issue that linux can't use FreeBSD pre-compiled stuff, and there is only partial support the other way around. So you are stuck having the entire docker image(all the way from and including the base image of Kali or other base linux distro) being recompiled every time you want to you run; which means you can't just "spin up" another docker image and have it working in minutes.
I am primarily talking about running FreeBSD container images. And no I would not recompile everything, just the layer that has changed, like with Linux/OCI containers.
Running Linux containers on this fictional FreeBSD jail-derived tech would probably work for quite some images, and would of course be cool, but is at least not my primary interest.
 
Even if, you do the ugly solution of putting if statements in the docker file, and detect which OS you are running, doesn't work in the end.
Docker is really just a distraction and a hinderance to actually solving this problem. Containers could perhaps be provided in source form or as you mentioned a more cross platform rules system to pull in the correct packages per platform.

Linux users and Docker Inc, are basically undermining this more portable development style. Oh well, at least it isn't Microsoft anymore. They were completely constricting.
 
prove would be some of my clients then, though that number is negligable. Probably even more companies at least could change their reproducible CI/CD toolchain and also their server fleet to FreeBSD then => more users = more feedback/quality improvement and finally more software vendors stop ignoring FreeBSD. Companies starting over would at least have a choice and not be forced to use Linux. It might not be what the majority of this discussion want, but there are some users like me who do.
Then fund it.;)
Seriously, if it's such a benefit, ante up the money and I'm sure someone will do it.

I'm not attacking your stance, you're entitled to your view and perspective, BUT, I find the idea of chasing corporate clients by picking up that latest thing from Linux (and which it already dominates anyway), a fruitless, pointless and costly exercise.

But, hell, if you want to lobby the FreeBSD foundation and the Linux foundation (which sponsors this OCI - of course, because it's on Linux), then I will try it out in FreeBSD 15. ;)
 
Focker looks interesting, I've played with it sometime ago, I'll surely take a look a 2.0, but for the folks that want some sort of native docker on FreeBSD, I think your asking in the wrong place.

IMHO and as many said it would take a huge investment from the Fundation into something where Linux already rules and will continue todo so, in the Clould world Kubernets is in fact very very interesting it become more and more mature and with would be a serious investment because docker and K8s are built from Linux.

It's been sometime since I use FreeBSD for desktop/laptop daily but I do see the usefulness of being able to run docker images in Freebsd, but tbh isn't that something to ask from Docker not FreeBSD.
I work mainly on macOS and docker implementation there use a VM, and in fact these this its actually using a port o xhyve which guess what? is in turn a port of bhyve.
All the Docker would need to do if there's enough interest (which I don't think there is) is to build a UI that works on any Unix, or at most have the cmd line be able to do what it's doing in macOS and just "pass all the cmds to the VM".

But for now what we can do is to simply implement that ourselves... If I wanted/needed to do work with docker and FreeBSD was my main desktop I would just setup a bhyve VM.... again this is exactly how it works on macOS and instead of complaining to Apple people are quite happy there's at least a way.
 
again this is exactly how it works on macOS and instead of complaining to Apple people are quite happy there's at least a way.
Yep and until Microsoft's semi-recent EEE strategy to put WSL on Windows, the official Docker Win32 installer would install VirtualBox and run Linux on there too. People were happy with that (FreeBSD also has VirtualBox in ports).

Docker isn't a portable standard or even a program which many people seem to think. It is basically just Linux automation. And people choose not to use Linux for many reasons.
 
For all those who wish Docker ran on FreeBSD and praise Linux for having such a thing--and the reasons I don't think this is off topic, here is another in a long list of reasons to not even use Linux at all--and we see such things too often as in Windows, too.

Goodbye Docker and Thanks for All the Fish

Docker has adopted Kubernetes. RedHat has dumped Docker altogether.

Yeah, I'm in a Linux bashing mood. I was reading some posts--on reddit (where else?)--where four people spent three paragraphs each praising BSD then stated "But I would never use it".

I know this is so stale, but FYI your entire premise was faulty. Kubernetes uses Docker, Docker didn't adopt Kubernetes.

I'm sure by now you have realized Docker is doing just fine.
 
I know this is so stale, but FYI your entire premise was faulty.

Not my article. Nor did I write any of the many articles online about Docker's demise. But I would never stoop so low as to use Linux so it doesn't affect me and I don't bother to follow such things anymore except if I stumble upon them.
 
From, Thread recap-of-first-meeting-of-the-freebsd-enterprise-working-group.90082/:

Cloud Native​

One attendee commented that of all the topics, Cloud Native is the most important in their view, perhaps more important than all the others combined. Allan brought a new project to the group’s attention, called XC: https://github.com/michael-yuji/xc Others asked how this relates to runj. The Open Container Initiative community is working on FreeBSD support and some Enterprise WG participants have expressed interest in helping. THANK YOU!
I recently learned about sysutils/podman which is available on FreeBSD and NetBSD. Podman has been mentioned on the forums a few times. For those who don't know, Podman is a container for Unix-like operating systems, and it's under the Open Container Initiative (OCI) standard. It's meant to be the equivalent of Docker, and compatible with its format. Podman is under Apache 2.0 License.

Relevant ports are sysutils/podman sysutils/buildah and sysutils/podman-suite.

OCI is a Linux Foundation project for an open standard for containers. I wonder why TrueNAS doesn't use Podman instead of Docker.

Edit: from that github link, xc is a container for FreeBSD 14. It is written in Rust and uses a BSD license.
 
Back
Top