Docker is dead

sadaszewski

Member

Reaction score: 12
Messages: 33

Well, yes. I most definitely think that it would!
However, from what I gather, it is about projects that I would be willing to work on myself. And I am VERY short on time :( I would like however to have a Docker-like solution for FreeBSD, either as evolution of Focker or as a Go-rewrite or perhaps C/C++ - rewrite if this is indeed what is the most desirable to the people. With added ability to use remote recipes as base in the build scripts and/or some sort of registry support (Docker-compatible or not).
 

suntzu00

Member

Reaction score: 11
Messages: 50

Don't you love when a fella comes around and takes things out of context?
Consul is a service discovery, connectivity, heart-beat, etc. tool. A much higher layer than Docker and certainly not a replacement or alternative
With a couple more tools like sysutils/consul and net/traefik you can get a kubernetes-like feeling
you should definitely ready my post again as I don't see where I mentioned consul and docker in the same sentence.

traefik seems to be more or less the same thing as consul.
wait, what?

cbsd is basically a wrapper of jail, jls and jexec
cheers man! I had no idea 👍

Doesn't seem to be FreeBSD/containerization friendly in this regard
sure they're not friendly. link link
 

vigole

Well-Known Member

Reaction score: 529
Messages: 445

466 days (bad chmod) and 6 pages. Apparently, Docker is not dead. He is in Jail.
 

sadaszewski

Member

Reaction score: 12
Messages: 33

you should definitely ready my post again as I don't see where I mentioned consul and docker in the same sentence.
Well, this is sort of a "Docker" thread. It is not unreasonable to expect that one would be trying to address the topic of Docker and alternatives, not propose unrelated solutions with completely different purposes, duuh.

wait, what?
Chill down, lest you get a heart attack ;) According to FreeBSD descriptions, consul is for "Service discovery and configuration" and traefik is a "High availability reverse proxy and load balancer". Close enough, they are definitely the same layer, way above container engines/Docker.

cheers man! I had no idea 👍
Again, this is sort of a "Docker" and/or containerization engine thread. One would expect adequate suggestions and it feels fine to point out things that are a little bit out of the whack.

sure they're not friendly. link link
Hehe, ok. It is something one should take a closer look at. Nevertheless it is only a 3rd party plugin to nomad. So we get to the situation that was previously ridiculed in this thread with respect to Linux/Docker, "front-end to this is a back-end to that" lol ;) So to get out of it a slightest resemblance to Docker (still without hierarchical images and compositions) you would need to put together Nomad, the jail-task-driver and pot. I think this is a strong argument in favor of Focker ;)
 

rafnizp

New Member


Messages: 3

Guys
You all are talking about pros and cons in regards to docker, kubernetes etc.
But I do not understand one simple thing. Where exactly is the problem of having those tools on BSD?
If there are people out there who wish to use this tools on BSD why we are limiting them?
Maybe there is some limitations on BSD that prevents from developing those solutions so if there is anyone who can explain more in details where is the problem I will really appreciate that.
 

rootbert

Well-Known Member

Reaction score: 103
Messages: 270

Where exactly is the problem of having those tools on BSD?
If there are people out there who wish to use this tools on BSD why we are limiting them?
There is no problem, there is just a lack of interest and/or human resource to develop an application container solution for BSD. Whoever wants or has to use such technology has already switched to a platform that provides that (e.g. Linux + OCI or smartos). I think we do not limit anyone, everybody is free to extend FreeBSD in a way one wants. Having an application container platform with BSD is an absolutely nice imagination for me (probably on top of my list), I could get rid of quite a lot of nasty Linux boxes. However, I am well aware of the tremendous amount of work that involves, so I won't even start without commitment of the FreeBSD Foundation. I hope though I will soon have time to have a look at Focker ... To be honest, the sad truth is, the size of the project is not necessarily an argument pro switching to BSD as an application container platform, even if requirements are met. The size of a project (numer of developers, commitment of other companies) is an important factor in management decisions.
 
OP
D

drhowarddrfine

Son of Beastie

Reaction score: 1,622
Messages: 3,666

Whoever wants or has to use such technology has already switched to a platform that provides that
This is a very good point I try to make to people. I don't use FreeBSD cause I want it to work just like Linux or Windows. I use FreeBSD because I find it superior to the others for what I do.
 

Jose

Well-Known Member

Reaction score: 227
Messages: 383

It may already be there with some Atheros cards. I haven't tried myself, but judging from the second to last posto at https://forums.freebsd.org/threads/intel-wireless-3160-poor-performance-speed.75733/#post-465837 some people are getting good speeds with it.
It is my understanding that the wireless stack in the kernel does not support 802.11ac. An adapter can claim to support it, but Freebsd won't use it.
The Foundation announced they're going to start sponsoring the work "soon" as of March 11, 2020. I'm no kernel hacker but I suspect it's going to take more than a few weeks for it to be completed and released.
 

rafnizp

New Member


Messages: 3

Correct me if I'm wrong, but from what I know BSD uses version control to their entire ecosystem and it is nothing like linux.
In my opinion if BDS can support most of those container technologies it can become very powerfull BSD CORE CLOUD OS
And this is what we think are missing out there this days, good quality system with well documented changes, fixes etc.
I think if BDS supports that many people will switch ower and build their private cloud based on this BSD CORE CLOUD OS
 

rafnizp

New Member


Messages: 3

Freebsd already has traction in the cloud: https://papers.freebsd.org/2019/fosdem/looney-netflix_and_freebsd/

What's missing is a daily driver for developers. Since programmers tend to be technophiles, many of them are not going to be happy with being stuck with 802.11n in 2020.
I do agree with you, that you will need to have fast communication (802.11ac) to build flexible and next gen applications (containerisation)
So both solutions I think will make huge impact on developers :)
 

vigole

Well-Known Member

Reaction score: 529
Messages: 445

Linux users have inflicted a pain on themselves by needlessly bloating software and breaking APIs
To be fair, they never said GNU/Linux is Unix (GNU's Not Unix)
people who think it's a good idea to have packages for the most trivial tasks which depends on 100s of packages
Do you have Latex in mind?
All thanks to a short-sighted vision of modern software development which claims to be "agile"
"Agile" in itself is a claim; offer a solution, write a book, search for the problem.
He said, get a job; I told him, read a copy of Structured Programming (Dijkstra, 1972).
These are some interesting articles/videos on DevOps
DevOps is just another marketing term
I had a professor who compared Docker and DevOps to the "Unix philosophy"
Send him/her a copy of FreeBSD Developers' Handbook | Architectural Guidelines
 

Hakaba

Active Member

Reaction score: 92
Messages: 187

I am pretty sure that poeple wants a "FreeBSD Docker" solution only accept the solution if Dockerfile and all docker cli option matches.
IMHO a Kubernetes "official port" with jails and/or Bhyve is a better answer (and maybe create a Jailfile and/or Bhyvefile to push any project inside a container).

Kubernetes for FreeBSD is not impossible
 

kpedersen

Daemon

Reaction score: 1,021
Messages: 2,020

Do you have Latex in mind?
Honestly all the packages used to be my least favorite thing about latex.

But then I realised that everyone pretty much just installs the *-full version of the distribution and rarely uses an external package other than the *one* hacked together by your intended journal or conference.

So for all intents and purposes, latex is absolutely monolithic and many gigabytes to install haha.

The fact that many packages cause so many side effects and breakages is still an issue. But that is probably more a symptom of latex being evolved rather than designed I suppose ;)

Disclaimer: I work at Microsoft which owns Azure and npm (the latter via GitHub). I don't work on either team.
We won't hold that against you... too much ;)

Your mention of technical dept is so very true. We are going to end up with a lot of broken systems if we keep on the way we are. It will also happen fairly abruptly as these large central services start to disappear.

Obviously making developers (and thus the world) rely on them as part of their business model is the important part. I don't think they have much consideration for lifespan or maintainability and other very important areas of engineering.
 

sadaszewski

Member

Reaction score: 12
Messages: 33

I am pretty sure that poeple wants a "FreeBSD Docker" solution only accept the solution if Dockerfile and all docker cli option matches.
IMHO a Kubernetes "official port" with jails and/or Bhyve is a better answer (and maybe create a Jailfile and/or Bhyvefile to push any project inside a container).

Kubernetes for FreeBSD is not impossible
This is completely unjustified speculation. I am myself a user who does now want this and I know at least a dozen others who don't want that either. Number 1 priority is to have similar abstractions and convenience level to Docker. Focker provides that nicely.
 

Zirias

Aspiring Daemon

Reaction score: 463
Messages: 961

This is completely unjustified speculation.
No, it's a sane assumption, because for any other usecase, there are already working solutions with FreeBSD. I'm getting increasingly fed up with your aggressive marketing for an open-source(!) tool. Marketing as such is fine, but hijacking a lot of unrelated threads over and over again is not.
 

vigole

Well-Known Member

Reaction score: 529
Messages: 445

I'm getting increasingly fed up with your aggressive marketing for an open-source(!) tool
The account has created on May 19, 2020. The first post was posted, the same day, on May 19, 2020, and it was about Focker. We've been reading about Focker in this forums for last 45 days: a total of 28 posts, 25 posts about Focker, in 45 days.
Therefore, it is clearly a form of product advertising. But I don't think, the first clause of FreeBSD Forums Rules has been broken by the user. i.e. The FreeBSD Forums are not an advertising billboard. Because this rule is phrased in a rather ambiguous way.
I'm not forum moderator, hence I'm only able to share my observations.
 

Zirias

Aspiring Daemon

Reaction score: 463
Messages: 961

It's unlikely there is any financial benefit in advertising an opensource project, so I wouldn't expect this to apply (but it's not my role to interpret the rules...) Still, it's just getting on my (and, obiously, some others') nerves ;)
 
Top