Important announcement regarding FreeNAS Corral

[johan56]

So Docker is gone?

 

[Michael Schultz]

So Docker is gone?

Yup, and won't be back for ~3 months.

 

[jpringle03]

FWIW, the devs are open to having a migration path from Corral to 9.10.3, and I'll be sure to lobby for that to happen. Docker is the only complication, so let me ask something: Is it better to hang on to Corral until a 9.10.4 that includes Docker support or migrate the Docker VM to 9.10.3, despite the lack of Docker integration with the host?

I think we've all agreed on what's wrong, so let's discuss how to fix things.

I think there should be options in both:

9.10.3 should have a migration path that preserves the Corral VMs but the containers themselves don't work anymore. I do see a task about making an easy way to do some Docker-ish like stuff in their bug tracker: https://bugs.freenas.org/issues/23357

9.10.4(or something) should have a migration strategy for VMs AND containers.

I did create this task: https://bugs.freenas.org/issues/23362

 

[bestboy]

They don't have much of a choice. They took away absolute required features and now people can't use freenas because it lacks functionality. If they want people to stop jumping ship they need to get a product with similar features to corral as soon as humanely possible. All they did with the release of corral was show their 9.* audience how many features they lack. If corral doesn't have full docker/vm support and a completed new gui by june, freenas is done for.

Michael I believe you misunderstood my point. I'm not against back porting Corral features. I'm just against doing it in a point release, which is merely nomenclature.

In software engineering we often use versions separated in 3 sections by dots (like FreeNAS 9.10.2)
The first section (9) is the major version. The major version is incremented, when there are fundamental changes (e.g. FreeNAS 8 vs. FreeNAS 9)
The second section (10) is the minor version. The minor version is incremented, when there are improvements to existing features or small new features.
The third section (2) is the maintenance version. The maintenance version is incremented, when a bugfix or a security patch is delivered without changing the feature set at all.

A point release is a release with a new maintenance version. Point releases should be applicable without any concern of breaking things or backward compatibility. Point release updates will therefore be applied more or less blindly by admins.
In contrast to that, major or minor version updates will always result in a special and careful treatment. Such updates require preparation, testing, maybe even migration. They will not be applied willy-nilly and often also require a recovery plan for when things go south.

So for me, releasing such profound changes as the ones from the back port of Corral as a point release to 9.10 (9.10.3?) is just wrong.

 

[johan56]

Yup, and won't be back for ~3 months.

Alright, thanks!

Can someone ELI5 what the thing is with Jails and Docker?

I have been tempering with Corral a few hours, and reading up on Docker, and Docker seems to be awesome! If I have understood things right, Docker ships apps in container and that is essential Virtual Machines? And there is a huge network of existing applications for Docker? (https://hub.docker.com)

Is this to be removed in 9.10.3?

Because I was really looking forward InvoicePlane =)

 

[Stanri010]

FWIW, the devs are open to having a migration path from Corral to 9.10.3, and I'll be sure to lobby for that to happen. Docker is the only complication, so let me ask something: Is it better to hang on to Corral until a 9.10.4 that includes Docker support or migrate the Docker VM to 9.10.3, despite the lack of Docker integration with the host?

Are you saying that when they push the new update, they won't have docker support upon the initial release? The only reason I upgraded and stuck with corral was for docker support to get away from the old jail system. Is docker plugin compatibility a priority for the new release or is it going to be just a new feature update down the road. I refuse to move to the new version until full docker support has been implemented.

The issue that FreeNAS is going to face now is Corral, while not perfect, exposed many of the issues with the old FreeNAS implantation. Going back to 9.10.x means we'll lose out features such as parallel command execution right?

• New Angular-based web UI: You can test-drive the early work now in 9.10 nightlies prior to the upcoming 9.10.3 release.

• Expand and improve support for jails and jail-based plugins: For maximum compatibility with lighter system requirements.
  
  
• VM Support: We have added a new “VM” menu which allows you to host your own Virtual Machines on FreeNAS, landing in 9.10.3.

• Docker support: As a Virtual Machine-driven service.
  
  
• Improve support for DevOps-class alerting, PagerDuty, AWS Alerts, OpsGenie, and Slack (coming in 9.10.3).
  
  
• Local and distributed S3 bucket support: Initial work landing in 9.10.3.
  
  
• FreeBSD 11-stable base: Landing in 9.10.3.

 

[NAStyBox]

And sometimes trying to piecemeal updates onto existing versions isn't the answer either. You can end up with Frankenstein code that is so pieced together that competitors can come along with fresh code bases that run MUCH smoother. I'm just saying, there are pros and cons to this approach that I'm sure the team weighed, but for myself and all of the people that are going to have to do a ton of work to resolve this disaster, I'd rather we not suggest that this is 100% the best idea. None of us know if it is.

I'm not trying to be condescending, although I know it may come off that way, but you're under 30 years of age, or have less than 5 years in the IT business, correct?

You never, and I mean never put a major release into production within 30 days of its release. In fact most large enterprises will test a minor release for 30 days. You may weigh the pros and cons, and sure with some desktop applications it's fine to jump in a little sooner, but that's not what this is. You're dealing with a piece of code that manages data integrity. This is the type of thing that any competent change manager would push back 3 months on the calendar, at a minimum, and that's only if the current solution is problematic and exhibiting high severity issues. In an enterprise environment it would normally take at least 6 months of reviewing bug reports and minor releases to even entertain putting it into production. It would certainly have to be running in a dev environment for some length of time.

So while it's a shame that Corral isn't going to make it to prime-time anytime soon, unless you're just a hobbyist you need to own this as your failure in practice. The folks at IXS did their job. They weighed the pros and cons of continuing the code stream as a stable release, and decided it wasn't ready. That almost never happens. Usually a software company will send their customers on a Donner party expedition, using a bad GA release as a test bed for the next major that might actually work.

Really. This is a gift. IXS probably protected the sanity of many of us, lol.

 

[Jeroen Sen]

FWIW, the devs are open to having a migration path from Corral to 9.10.3, and I'll be sure to lobby for that to happen. Docker is the only complication, so let me ask something: Is it better to hang on to Corral until a 9.10.4 that includes Docker support or migrate the Docker VM to 9.10.3, despite the lack of Docker integration with the host?.

+1 to hang on to Corral until a 9.10.4 that includes Docker support

 

[Stanri010]

If 9.10.4 is the version that gets docker support, will be able to point our new instance of the plugin on 9.10.4 at the old configuration folders and have it pick up right where we left off?

I really don't want to lose my Plex library and the way it's currently set up...

 

[Michael Schultz]

FWIW, the devs are open to having a migration path from Corral to 9.10.3, and I'll be sure to lobby for that to happen. Docker is the only complication, so let me ask something: Is it better to hang on to Corral until a 9.10.4 that includes Docker support or migrate the Docker VM to 9.10.3, despite the lack of Docker integration with the host?

I think we've all agreed on what's wrong, so let's discuss how to fix things.

I am going to respond to this but I want to calm down first. I am going to go on a 4 hour bike ride because I requested the day off 3 weeks ago to finish my FreeNAS Corral installation and that obviously doesn't need to happen... I wish that was a joke.

Alright, thanks!

Can someone ELI5 what the thing is with Jails and Docker?

I have been tempering with Corral a few hours, and reading up on Docker, and Docker seems to be awesome! If I have understood things right, Docker ships apps in container and that is essential Virtual Machines? And there is a huge network of existing applications for Docker? (https://hub.docker.com)

Is this to be removed in 9.10.3?

Because I was really looking forward InvoicePlane =)

Docker is amazing because it takes system configuration out of the equation when deploying code. The software requirements come with the docker "package" so you are never missing anything. Need to move from rhl to deb because tony's is a dick will only apt-get? Super easy, just move the docker container over, import it and your done.

That is why I am not all that concerned with an upgrade path from corral to 9.10.*. The whole point of docker is that i should be able to just "# docker import" and be up and running again. I'll have to remap disk/ports/env variables but if that's all I have to do that's easy.

I guess what I am really trying to say is I don't want to use dead software for the next 3 months while docker functionality is added, but unfortunately that seems unavoidable.

 

[Kris Moore]

Unfortunately, it doesn't sound like iX has learned about release engineering. A point release (9.10.2 to 9.10.3) is simply not an appropriate place to introduce a whole new GUI, just as a service update to a point release is not an appropriate place to remove a feature that lots of your users use (like VirtualBox).

The 9.10 branch has always had .X releases as "feature" releases, the -UX releases are patch/bugfix updates only. That being said, the reason for that versioning scheme was political while Corral development was underway and we will be revisiting it here in the near future. (Believe me, we like it less than you do)

 

[Middge]

FWIW, the devs are open to having a migration path from Corral to 9.10.3, and I'll be sure to lobby for that to happen. Docker is the only complication, so let me ask something: Is it better to hang on to Corral until a 9.10.4 that includes Docker support or migrate the Docker VM to 9.10.3, despite the lack of Docker integration with the host?

I think we've all agreed on what's wrong, so let's discuss how to fix things.

Until full support for Dockers and VM's comes to 9.x.x, I will continue using Corral. That includes full migration support.

If 100% point to point migration is never implemented, I will definitely be leaving FreeNAS for good.

This probably sounds petulant, but the amount of hours of free time I have invested into getting Corral working the way I want was such a high price to pay that I will not be paying it again for the same product.

I seriously love FreeNAS. I even loved 9.10, but literally stranding us on an island of abandoned ideas was a bad play. I really hope you can provide migration options because in my honest opinion that is the only way to recover from this.

 

[Magnus33]

Not everything is doom and gloom.

Corral was doomed from the start due to some choices made in its development but the ideas behind it where actually will thought out.
The ui in version 9's of FreeNAS is and has been a convoluted mess or basically something made by engineers for engineers which has been happening since software development began lol.

The Corral ui kept controls to one side and gave people options for much more information that been seriously lacking in version 9 (seriously lacking basic info like hardware temperature really!!)

Jails though quite stable is also at a dead end for ages now.

Development has been shooting itself in the foot for ages not to mention keeping interested investors away.
You can't run a bug reporting system the way its been done which is for lack of a better word a free for all and not affect your long term development and support.

Its great when you doing a project with a group of friends but when in terms of a business its a bad idea.
Linksys never hit main stream for this very reason.

Ubuntu shows that you can have the open source development with a more structured control to prevent the too many cooks in the kitchen issues that FreeNAS developments always had.

 

[Ericloewe]

Going back to 9.10.x means we'll lose out features such as parallel command execution right?

Not long-term, I'll probably write up a features FAQ to go along with the migration one.

 

[Magnus33]

Until full support for Dockers and VM's comes to 9.x.x, I will continue using Corral. That includes full migration support.

If 100% point to point migration is never implemented, I will definitely be leaving Freenas for good.

This probably sounds petulant, but the amount of hours of free time I have invested into getting Corral working the way I want was such a high price to pay that I will not be paying it again for the same product.

I seriously love Freenas. I even loved 9.10, but literally stranding us on an island of abandoned ideas was a bad play. I really hope you can provide migration options because in my honest opinion that is the only way to recover from this.

I doubt 100% migration is going to happen its just not a priority.

Dockers on the other hand shall have to happen since jails have been at a dead end for years now.

But yes itx has some serious choices to make at this point which is likely going to affect the future of freenas.

If there had been more structure in development and bug reporting corral would likely never have head done the code choices it did putting it in the place it is.

 

[nukee]

When should we expect the forums to be restructured back to what they were pre-Corral? Kinda useless having a dead release front-and-center.

 

[Magnus33]

One thing the team must understand is this is going to have a long term impact on the company.

By suddenly reverting after a month they spooked the hell out of a lot of people and that not good in business.

Great features like bhyve VMs and Docker containers are for now gone unless you stick with the corral preview.

FreeNAS is still the best NAS storage system but the future is now very questionable on if its going to remain so and frankly for the company too.

 

[Ericloewe]

Great features like bhyve VMs

Bhyve is present right now. Bhyve integration in the UI is coming with 9.10.3.

 

[Ericloewe]

When should we expect the forums to be restructured back to what they were pre-Corral? Kinda useless having a dead release front-and-center.

We're discussing this. The current layout was temporary while Corral established itself, so plans have to be revised, servers snapshotted to prevent mystery data losses, etc.

 

[Diosyntaxa]

Not everything is doom and gloom.

Corral was doomed from the start due to some choices made in its development but the ideas behind it where actually will thought out.
The ui in version 9's of freenas is and has been a convoluted mess or basically something made by engineers for engineers which has been happening since software development began lol.

The corral ui kept controls to one side and gave people options for much more information that been seriously lacking in version 9 (seriously lacking basic info like hardware temperature really!!)

THIS! So true. I'm confident that it will all be sorted in due time though. The new-new UI looks a bit disappointing to be honest, but it looks a bit clearer at first glance than 9.x at least, and assuming it's not painted into a corner, things can, and should change for the better with time.

I'll stick with you guys through this, because from the first time I realized the strengths of FreeNAS I've been a fan of it. There have been ups and downs, and I really hated the old GUI, but hopefully things will be learned from this, and from the feedback received about what was good about Corral. For me this has been the GUI, CLI, dockers and simpler support for VMs (despite not actually doing much with this, it's one of the reasons I got a server that would be total overkill for just a NAS) and the overall structure of it all with everything being more easy to take in.