Docker is dead

They provide some simple skeletons you do not really need to know OCaml to use. Also, if you do some search you will find some others people constructed and made available. I saw some time ago some people willing to construct one for Ocsigen.

[EDIT]

The Muen is a particular interesting case. The site run on a MirageOS unikernel virtualized on top of Muen.
 
rigoletto@ said:
I would rather prefer something properly developed like unikernels instead. MirageOS is indeed a good example. And, for the record, MirageOS work on bhyve.
Click to expand...
That is yours. Not the problem I see. I said about traditional OSes. Prebuilt images of something like CentOS or Ubuntu already made by someone I could fetch and modify to suite my needs.
 
badbrain said:
Straightforward. I like it. If we go with this answer sooner no useless off-topic discussion needed. Back to the topic about docker.
Click to expand...

Just because a company has not yet added an almost consumer-centric web page to a technology, does in no way mean it is inferior.

Something like DockerHub is great for quick thrills and if someone is unable to install complex software themselves but in real world scenarios, most of the images are not usable; they are not tailored to the job at hand. I would still need to set one up bespoke for any internal developed software.

Do you think people just use Git because of Microsoft GitHub? Why do people still also use CVS, SVN, Hg, etc.. when there are no corresponding *Hub sites for them?

Don't get me wrong, containers are good and Docker is fine. But I find the off-the-shelf container "market place" design a little naive especially when it comes to future maintenance.
No-one is going to maintain a i.e glibc 2.11 container just so I can run some crusty old application. I will either have to update my stuff or maintain the container myself. So I might as well start now and ignore the DockerHub.

Docker is one of those things that is great until it all stops working in 10 years. Then we will need to dig up all the old shite and replace it. Yeah, we will get paid, so I can't complain I suppose.
 
kpedersen said:
Just because a company has not yet added an almost consumer-centric web page to a technology, does in no way mean it is inferior.

Something like DockerHub is great for quick thrills and if someone is unable to install complex software themselves but in real world scenarios, most of the images are not usable; they are not tailored to the job at hand. I would still need to set one up bespoke for any internal developed software.

Do you think people just use Git because of Microsoft GitHub? Why do people still also use CVS, SVN, Hg, etc.. when there are no corresponding *Hub sites for them?

Don't get me wrong, containers are good and Docker is fine. But I find the off-the-shelf container "market place" design a little naive especially when it comes to future maintenance.
No-one is going to maintain a i.e glibc 2.11 container just so I can run some crusty old application. I will either have to update my stuff or maintain the container myself. So I might as well start now and ignore the DockerHub.

Docker is one of those things that is great until it all stops working in 10 years. Then we will need to dig up all the old shite and replace it. Yeah, we will get paid, so I can't complain I suppose.
Click to expand...
Such thing like Dockerhub is so helpful for outsourcer. Specifically for devs on our country. They are pretty much just about money, deploy fast and gone, despite the code is spaghetti and shitetty because another random guy of another low cost outsource company will take that shite. Somewhat irresponsible but true. Even their cash is too low compared to the western. And the company keeps most of it.
 
badbrain one thing you didn't got is FreeBSD does not support crap solutions, we prefer to be slow and do it technically right than offer crap and that is, as expected, the way our audience like the things done. Also, dominate the market like Linux is not one of our goals, inclusive because that would involve making enormous amount of compromises to fit "everyone" taste and necessities, which inevitably translate to low quality.

What we are failing is to make that clear marketing-wise, but this is something the we are working on while we talk, in several different ways.
 
What I always wonder is why people do not just write Makefile-like scripts to allow developers to setup a clean environment / easily install an application and all its dependencies and configuration in one click. This would anyway be needed to have repeatable builds.

Regarding the argument: to avoid bloating personal computers, Linux has had LXC containers for quite a while now. So I think the real issue is that people are comfortable with crap engineering practices. They're comfortable loading a full OS in RAM while a shell script could have made it. People prefer to learn ever-changing layers of abstraction such as docker "just" so that they do not need to learn system administration which is supposedly too hard.

I even think that in most cases, a configuration management solution is a better solution than Kubernetes. If you need "cloud-grade scalability", all cloud providers sell transparent scaling as a selling point. If the goal is to easily move from on-premises to the cloud as you scale, a Makefile along with a configuration management solution gives you this result. If the idea is to move your 10TB database to the cloud, you'll need to ship your hard drives anyway unless you can wait 6 months for the upload to complete.

I kind of understand why people choose to go the lazy way, but I find it very unreasonable to waste so much resources. But the good thing is that these people keep cloud providers in business so that those of us investing in sound engineering (which is not such a huge investment, only requires goodwill and consciousness) can run multillion-dollar businesses having hundreds of sandboxed applications running on a single box for 200 dollars a month while others make cloud providers rich.

I think if you are running FreeBSD, considering the power of jails and Linux emulation, you really do not need Kubernetes or Docker. Just write setup and deployment scripts. If people put the effort they invested in Docker into learning FreeBSD (which has been providing a great environment for these conveniences for a very long-time, jails, linux compatibility layer, the port system, zfs etc...), the world would probably be a better place. I am not even a FreeBSD wizard but the system is easy to use, stable, well designed and just works.
 
Here we go again. 100 people with used laptops, open source idea, leased office - valuation 1 billion :)
Who's doing all the valutions of those "We got the killer app or solution," and where do all the billions, in IPOs and adverts, are coming from?
Bill Gates invests in his own fundations, Paul Allen blew his money to look for radio signals from outer space, Elon Musk and Jeff Bezos are blowing their funny money on rockets, Google is giving money away to YouTube stars, Warren Buffet doesn't play High Tech ...
I think the money changers are creating money out of thin air and are giving it away to anyone with a good or bad idea and a laptop - LOL
 
toorski said:
Here we go again. 100 people with used laptops, open source idea, leased office - valuation 1 billion :)
Who's doing all the valutions of those "We got the killer app or solution," and where do all the billions, in IPOs and adverts, are coming from?
Click to expand...
For example, it comes from hype companies like facebook. These are highly depending on their share price. So what would happen if Zuckerberg was selling half his shares? The rest would crash down. But if he buys a startup with it that was funded by "a guy whose 3.rd cousin he met in a pub"? Vision! Progress!! In the end he moved a lot of his wealth from fantasy land (shares) into the real world.
 
drhowarddrfine said:
Not to repeat the title of this thread but ...
Click to expand...

This is great. Just like Adobe Flash, Unity, DotNet and a lot of other highly marketed products, it is useful when things go under; it should be a learning experience for developers to check the dependency and maintenance trail before investing their livelihood on the sole bucket of convenience.

But the reality is that they don't learn; these guys will instead become middle management and then a new batch of naive developers come up and jump on the naxt big flaky bandwagon. Together they continue making terrible technical decisions and software. They then fade away leaving the industry with no real progress.
 
wow. this is some negative circle jerk ya'll have going on here. Are there any mature FreeBSD leaders in this thread?
I think it's fine if BSD does not want to be docker compatible, there is lots of room in technology for different solutions. Mainframes even still have their place after all. However, wishing death upon another technology simply because it doesn't work well on your platform out of the box is rather naive.
With regards to those saying "bsd does things technically correct and the rest is crapware"....well, that's an interesting perspective given how many "technical facts" about docker in this thread are dead wrong. For instance, Docker does not load an entire OS into memory. Also, docker is keenly aware of LXC - it utilizes it! One could almost say that Docker simply created a friendly user interface for LXC. And that is just one example, I am actually hard pressed to find a technically accurate statement about docker in this thread.

From a non-technical perspective, there also seems to be big gap in understanding. I see some folks suggesting Make files - that's literally the anti-thisis to docker! Make files require an understanding of the platform the software runs one so as it can ensure it has the right dependencies, and with some luck, those dependencies won't conflict with other software on the same system. By in large, docker solves both of these very real problems and so many more.

This is my first foray into the FreeBSD community and if this thread is indicative of overall culture, yikes! As a professional sys admin, I am inclined to steer away quickly. Weather Docker succeeds or fails in BSD, the extremely uninformed hyperbolic statements made by it's members - none of it is professional in tone, purpose, or accuracy - are definitely not for me and my peers.

Good luck to ya'll !
 
somedude said:
statements made by it's members - none of it is professional in tone, purpose, or accuracy - are definitely not for me and my peers.
Click to expand...

If you judge the merit of a technology based on members of a forum then no... FreeBSD is definitely not for yourself or your peers.

But we are all very thankful that you took the time out of your day to sign up to these forums in order to revive a 2+ month old thread. We are all much more knowledgeable now ;).

Honestly though, I am surprised Docker is still even a thing in 2020.
 
somedude said:
I think it's fine if BSD does not want to be docker compatible
Click to expand...
It's not up to FreeBSD to do this. Docker is a Linux thing only. Some try to shoehorn it into FreeBSD but that's not what this thread is about. The "Docker is Dead" title comes from two articles from two non-FreeBSD sources but here you are pointing the finger at FreeBSD.
somedude said:
This is my first foray into the FreeBSD community and if this thread is indicative of overall culture, yikes!
Click to expand...
That's nothing. You should see any one thread in any one technical Linux forum. This is the Off Topic board here.
 
somedude said:
wow. this is some negative circle jerk ya'll have going on here. Are there any mature FreeBSD leaders in this thread?
I think it's fine if BSD does not want to be docker compatible, there is lots of room in technology for different solutions. Mainframes even still have their place after all. However, wishing death upon another technology simply because it doesn't work well on your platform out of the box is rather naive.
With regards to those saying "bsd does things technically correct and the rest is crapware"....well, that's an interesting perspective given how many "technical facts" about docker in this thread are dead wrong. For instance, Docker does not load an entire OS into memory. Also, docker is keenly aware of LXC - it utilizes it! One could almost say that Docker simply created a friendly user interface for LXC. And that is just one example, I am actually hard pressed to find a technically accurate statement about docker in this thread.

From a non-technical perspective, there also seems to be big gap in understanding. I see some folks suggesting Make files - that's literally the anti-thisis to docker! Make files require an understanding of the platform the software runs one so as it can ensure it has the right dependencies, and with some luck, those dependencies won't conflict with other software on the same system. By in large, docker solves both of these very real problems and so many more.

This is my first foray into the FreeBSD community and if this thread is indicative of overall culture, yikes! As a professional sys admin, I am inclined to steer away quickly. Weather Docker succeeds or fails in BSD, the extremely uninformed hyperbolic statements made by it's members - none of it is professional in tone, purpose, or accuracy - are definitely not for me and my peers.

Good luck to ya'll !
Click to expand...
As a professional sysadmin/programmer, i've never used docker, but then in my professional capacity i dont use linux.

Docker is a linux-ism and I do wonder about the logic of attempting to 'make it run' on FreeBSD. It seems like a huge waste of time; a bit like trying to get xfs to run natively on FreeBSD. Oh well, i guess how a porter or developer spends their time is up to them but boy what a waste of time attempting to port docker or kubernetes.
 
I would love to see a docker version on FreeBSD, specifically a jailed version - ie a docker/OCI container or a group of them in a jail with a nice CLI management tool on the host ... thats what I really dream about, and probably this would make FreeBSD a viable option for a range of things again. For all my largescale projects FreeBSD is of no interest due to lack of Docker. Providing a cloud system with 100 thousands of docker containers is so incredibly easy, all test driven from a nice CI/CD pipeline.

Having just a few lines of config/docker compose lines producing reproducable, testable containers is one thing I don't ever want to miss again, so imho developing this for FreeBSD would save loads of developer resources if you try to implement this with jails etc. - I think in the end it would pay off.
 
You must log in or register to reply here.
Back
Top