Important announcement regarding FreeNAS Corral

[danb35]

didn't docker run in a VM on Corral instead of native on Freebsd?

Yes, Docker is a Linux thing.

 

[Mlovelace]

How the hell did that ever get released?

That was my thought when the posts started rolling in about how to replace a failed drive and the response was CLI only. Just, WOW!

 

[gbooker]

How the hell did that ever get released?

The whole release of Corral struck me as quite sudden. I was regularly monitoring the bug reports attributed to (then) FreeNAS 10, and I noticed the bug list was still quite long just a few days before release. Clearly it was not ready in many aspects.

In retrospect, I have to wonder if the release was pushed out ahead of when it was ready due to some mandate from above. Given the number of people who've left, been forced out, etc, it seems that some sort of politics is the most logical explanation.

 

[danb35]

Wow indeed. I knew release management practices at iX were... unusual... but to knowingly release a "STABLE" NAS product without a GUI for disk replacement speaks to strikingly poor judgment on the part of whoever made the decision to release.

Kudos to iX for ultimately doing the right thing and killing it, I guess, but the more that comes out about it, the more it sounds like one inexcusably poor decision after another.

 

[Magnus33]

In general? Do you mean that some guy or some people with certain setups had issues with disk replacements? Or, nobody could do disk replacements at all?

Adding all of these features to 9.10 is going to create bugs "in general" also.



Having two systems that attempt to accomplish the same functionality in the same application is bloat.

Agreed jails are dead and have been dead for some time now.
Dockers add much for functionality, flexibility not to mention there far more support for them in the wild.

What does that overhead quantify out to? Are we designing the system to run on a computer that grandma gave away because she replaced it 5 years ago? The overhead difference of Docker is easily outweighed by the ubiquity of its use, community support and configuration model.

Agreed you have the weight the usability versions the usefulness and jails have no development behind them plus compared to dockers there very very few of them.



That is good news at least.

However, is that without bugs? When is the bug testing for that major overhaul to 9.10 starting and ending? Some people are claiming that Corral was released too soon without enough testing. Yet, we're talking about releasing significantly altered versions of 9.10 with basically the same changes that Corral offered at an even faster pace.

Of course the idea of bug free is nuts as there always bug in any software this is not something that can be avoided.
Hopefully they copy much of corral much improved ui and not keep going the way they are since the new ui is a mess.

The huge issue is the process that caused the issues in corrals development are all still there which is whats been keeping companies and people from investing in them.
There too many cooks in the kitchen and not enough control in place ..with a bug reporting system that's little better then a free for all.

Hell someone should have stepped up and said if you on corral stay put tell they get 9.10 updated enough that there wont be any downgrade issues.
They did mention there could be issues but ti clear not one actually stopped to test things before they made that announcement.

They badly need to put some controls in place in both the bug reporting, announcements and development.
Smart people but its got that all over the place Linux development which has kept it from going mainstream besides ubuntu.

 

[danb35]

...and I'm probably speaking from a position of ignorance here. After all, I'm not a dev, I've never been a dev, and I'll almost certainly never be a dev. But I thought release practices in the software industry were relatively standardized. You had major releases and minor versions. And then someone added point releases, but those were still fairly well-understood. Major changes to software happened in major releases, minor versions might add some smaller features and fix bugs, point releases were bug fixes only. Beta software was supposed to be feature-complete though probably buggy, and RCs were set in stone barring show-stopping bugs. It's fairly straightforward, at least from an end-user perspective.

AFAIK, iX has never operated like this wrt FreeNAS, adding to user confusion since the days of FreeNAS 8.x, and it got crazy with the 9.x releases. I count eleven 9.2.x releases on download.freenas.org. Then we went to 9.3, changing the boot filesystem to a ZFS pool, adding mirroring, adding boot environments, and a bunch of other stuff (which probably warranted a major release), and the version numbering changed to 9.3-yyyymmddhhmmss, which results in pretty long version numbers, and gives no idea how significant (or not) any particular release is. One of those yyyymmddhhmmss updates got designated 9.3.1, even though it never displayed that way in the product itself.

Then we had 9.10, another minor release according to the version number, which involved a wholesale replacement of the underlying operating system--this certainly warranted a major release. Along the way, it broke everyone's jails, and we were told that the only thing to do was to trash them all and set them up from scratch (we were later told that this guidance was in error, but by then it was too late). It also confused plenty of newbies who correctly saw 9.10 as being numerically identical to 9.1, and therefore believed that 9.3 was a newer version. We went back to version numbering of major.minor.point, not using a fourth designator as was done in the 9.2 days, but adding -Un. From 9.10.1 to 9.10.1-U1, we deliberately and without warning killed VirtualBox jails. They were added back in -U2, but then permanently removed in 9.10.2, and no method of migration has yet been provided for those VMs.

Then we had FreeNAS 10. A few months ago, it was released as what we were told was a beta, though it was nowhere close to feature-complete. It made it to RELEASE status lacking critical features.

So, @Kris Moore, what's going on? Is it just that you (collectively) can't smack down the Good Idea Fairy? You say "Believe me, we like it less than you do," and that's good, I guess. So what are you going to do about it?

 

[MisterE2002]

Corral should never been released as STABLE. I think the GUI/Architecture is 100 times better than 9.x series. It is just a rough diamond.
Stepping up and admitting guild is a bold and correct move.

but...just firing the creators is a real disgrace.

The Corral team was presented by developers of many countries. The 9.x is the old (local) guard?
It seems the old guard never commited anything to this project. Now they seem to have the "Not Invented Here" syndrome.

Somewhere i read: "The whole middleware does not have a single functional test. Thats right, whole middleware was built without a testing harness."
In the 9.x commits i never seen a UnitTest commit. But maybe i am wrong.

I found all other arguments not that strong.
Now this whole company gives me a bad taste in the mouth :(

 

[William Grzybowski]

Corral should never been released as STABLE. I think the GUI/Architecture is 100 times better than 9.x series. It is just a rough diamond.
Stepping up and admitting guild is a bold and correct move.

but...just firing the creators is a real disgrace.

The Corral team was presented by developers of many countries. The 9.x is the old (local) guard?
It seems the old guard never commited anything to this project. Now they seem to have the "Not Invented Here" syndrome.

Somewhere i read: "The whole middleware does not have a single functional test. Thats right, whole middleware was built without a testing harness."
In the 9.x commits i never seen a UnitTest commit. But maybe i am wrong.

I found all other arguments not that strong.
Now this whole company gives me a bad taste in the mouth :(

There are (was) several remote developers for both projects.

9.x has Jenkins in place with over 550 tests being run for every commit.

EDIT: Yes, I have several commits to Corral project. I assure you the "Not Invented Here" syndrome does not exist.

 

[laskeyp]

7 was EOL before 8 ever existed. It also wasn't iXsystems.

I have been running 7 in a VM on ESXi since it was released and it has been rock solid for years. As a NAS system it did the job it was intended to do and did it well.

 

[adrianwi]

Wow. Feel for all those involved with Corral (aka FreeNAS 10) whether from a development perspective or early adopter.

Given '10' seems to be an unlucky number for iXSystems (could this have something to do with the iX part?) would it not be better to go back to 9.3.1 where jails (other than messages about being out of support when updating them) and VirtualBox are still running just fine?

It would also remove any potential headache for me in terms of which upgrade path to follow, or not :)

 

[NeedNAS]

I've been testing FN for several month to use in my companies IT system. I've played with FN9.10 and FN10 (Corral, I can't stand the rebranded name) since its the earlier stages, months before the "release". I'm pretty disappointed that that Corral is getting shelved, BUT I'm glad it happened before I started moving live data and VMs over.
I found out that Corral was being benched through a news feed, only after searching why I hadn't received any "nightly's" in a few days. One thing iXsystems should do is some how push an update letting people know the what is happening.

 

[KLEPTOROTH]

tl;dr: FreeNAS Corral as it was originally released is being relegated to “TECHNOLOGY PREVIEW” status while we work hard to re-base its exciting new features upon the rock-solid FreeNAS 9.10 base.


As many of you diehard FreeNAS® users know, we released FreeNAS Corral on March 15th, and the initial Community response was largely positive. There was a lot of excitement around the updated UI and the VM/Docker support, especially. However, we’ve also seen nearly half of the initial users revert back to FreeNAS 9.10. User feedback about this drop-off has been clear: challenges upgrading from 9.10, general instability, lack of feature parity with 9.10 (Jails, iSCSI, etc), and some users experiencing lower performance than expected given the increased demands FreeNAS Corral has on system hardware resources. With the subsequent departure of the FreeNAS Corral project lead, we re-examined the features, benefits, and issues with Corral and have decided to revise our plan for its future.

Before we communicate this new plan of record, a little background is probably in order. As some of you may know, the FreeNAS Corral GUI was built on MontageJS framework, originally working alongside the team at Montage Studios. Unfortunately, during the development of the product, the Montage Studio team disbanded, and the development of the MontageJS framework slowed to a crawl (this explains some of the browser incompatibility we’ve seen). So, our first goal following the release was to begin remaking the FreeNAS Corral UI (yes, yet again! …. /sigh) by basing the same UX on a more common framework. Not a huge deal, really, just some extra UI work for the team, but this time with a more common framework, allowing for faster development and more opportunity for contribution from the community. Once that new framework was in place for the UI, the next phase was to begin merging the FreeNAS 9 and FreeNAS Corral code bases and Engineering Teams.

However, in response to the volume of mixed feedback from the user community since release, we decided to undergo a thorough engineering review of the product and started to look deeper into the Plan 9 filesystem code, which allows VMs to access the host’s filesystem. In doing so, we discovered some holes in the architecture which make enterprise-quality file access using 9pfs impossible without a lot more effort and soak time, prompting us to to also re-think how to more safely enable this capability.

After weighing community feedback, and much internal deliberation at iX, we have decided that the amount of work still required to bring FreeNAS Corral (as currently architected) up to an acceptable standard for quality, reliability, and data integrity will take an unreasonable amount of time. The quicker path to a properly stable and enterprise-worthy Corral is to rebase upon the solid FreeNAS 9.10 code, bringing some of the new features that the current FreeNAS Corral offers into a more mature and solid platform. This process has already begun with the inclusion of VM container support and a brand-new Angular-based UI which is already available in the 9.10 nightlies (more on this below).

For the time being, the current release of FreeNAS Corral will be treated as an experimental branch and repositioned from “RELEASE” to “TECHNOLOGY PREVIEW” status, available for download and experimentation by the adventurous among you, but not for use in production environments. This also means it is unlikely you will be able to migrate configuration settings from Corral -> the next FreeNAS Corral product (however, your data will always be importable).

This new direction will allow us to focus our efforts on our next release which will merge the legendary stability of FreeNAS 9 with the whiz-bang features of FreeNAS Corral, while also swapping the GUI with the new Typescript framework. This provides the best of both worlds (stability + features) and has the added benefit of being a far faster path to a rock solid and stable FreeNAS release. In fact, many of the original team behind FreeNAS have already begun the process of taking the 9 series and merging it with some of the new features introduced in FreeNAS Corral, for the next stable and soon-to-be-released FreeNAS Corral.

In the meantime, our next release, FreeNAS 9.10.3 is currently slated for May, and here is a look at the current roadmap (subject to change as we move farther along, of course):

• 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.

Most of these items are already under active development, and we at iX look forward to sharing more details as they become available. As usual, we ask our beta-testing community to test drive these features in the nightlies and provide feedback and bug reports on the official tracker.

Thank you for your continued support and usage of FreeNAS. We appreciate all the users and fans who make this product better on a daily basis.

On behalf of the iX engineering team,


Kris Moore
Director of Engineering
iXsystems

Mod note:

There's an FAQ about moving from FreeNAS Corral to FreeNAS 9.10.2 in the Resources Section. You can find it at this link.

- Ericloewe

In comparison to the original Release of Corral, I think the new UI is bland. I LOVE the UI in the original FreeNAS Corral.

I hope you can find a way to skin / theme it so it looks the same. The left & right panes with the sliders when you select something in combination with the color scheme is amazing.

Don't get me wrong the new UI is nice (compared to the old v9 UI) had I not seen the OG Corral UI. But it just looks like an updated v9 UI, which I'm assuming is exactly what it is.

What are everyone else's thoughts about the UI?

 

[William Grzybowski]

In comparison to the original Release of Corral, I think the new UI is bland. I LOVE the UI in the original FreeNAS Corral.

I hope you can find a way to skin / theme it so it looks the same. The left & right panes with the sliders when you select something in combination with the color scheme is amazing.

Don't get me wrong the new UI is nice (compared to the old v9 UI) had I not seen the OG Corral UI. But it just looks like an updated v9 UI, which I'm assuming is exactly what it is.

What are everyone else's thoughts about the UI?

Keep in my new UI as far from finished. Its just a starting point (it was even based of an existing angular dashboard). A lot is still going to change (for the better).

 

[laskeyp]

Corral should never been released as STABLE. I think the GUI/Architecture is 100 times better than 9.x series. It is just a rough diamond.
Stepping up and admitting guild is a bold and correct move.

but...just firing the creators is a real disgrace.

The Corral team was presented by developers of many countries. The 9.x is the old (local) guard?
It seems the old guard never commited anything to this project. Now they seem to have the "Not Invented Here" syndrome.

Somewhere i read: "The whole middleware does not have a single functional test. Thats right, whole middleware was built without a testing harness."
In the 9.x commits i never seen a UnitTest commit. But maybe i am wrong.

I found all other arguments not that strong.
Now this whole company gives me a bad taste in the mouth :(

I agree. If Corral had been released as a beta or Technology Preview I would be writing about an exciting and very promising piece of new software, though it still has a few bugs. Instead, it was released as stable which it clearly was not. This was a management issue, not a technical one. Getting rid of the developers that worked on this seems to be yet another management failure.

 

[joeschmuck]

Folks,
I know that everyone has a complaint about Corral and how it got released and several other conserns, so did I, but this was an announcement thread. I recommend that if there are other complaints that a new thread be created. But my best advice is that when the next version of FreeNAS comes out, whatever that may be, you either test it on a seperate system or wait a few weeks and see what happens on the forums. There are always going to be early adopters and if there are problems, they will show up on the forums. And I don't think the developers will make that same mistake anytime soon of rushing a product out the door prematurely. Well I hope not.;)

 

[ChrisFitzFor]

Why was it ever moved to being a Stable release is my question? It was clearly nowhere near being stable.

Good question, probably that's why a couple of people had to go.

 

[mow4cash]

I found out that Corral was being benched through a news feed, only after searching why I hadn't received any "nightly's" in a few days. One thing iXsystems should do is some how push an update letting people know the what is happening.

I posted on this thread a week ago saying the same thing. At the time they still had Corral available for download with no warning(unbelievable) and no posts in the Corral forum. People are still submitting bugs for Corral. I don't understand why a notification update is not being pushed.

 

[maydo]

i see there is a corral 10.0.5 iso online, what about this version ?
is there any fixes ?

 

[indivision]

The idea of docker is great, but correct me if I am wrong but didn't docker run in a VM on Corral instead of native on Freebsd? To me this kind of makes docker containers a waste in most situations, aside from initial ease of setup.

To me, it gave the advantage of being able to set up multiple hosts that group containers with tailored server resources.

This...

And it wasn't in the GUI at all. Crazy, right?

I wouldn't call it crazy. That wouldn't be much UI work to add. And the functionality was available in the CLI for people who really needed it right away.

Bug in new features are one thing. Bugs everywhere because so much was rewritten is a whole different thing.

Yes. But, there comes a time in most software life-cycles when the changes that really need to happen can only be achieved with major re-writes. And a larger list of bugs goes with that territory.

We also have several types of programming languages and yet you can't say that C is bloat because there's Python available.

That is apples to oranges. A software application doesn't compare to programming languages in general.

Surprisingly nasty for the common Plex server scenario, it seems.

I don't see how that is possible at all.

I have a relatively modest machine and I was limiting Plex' resources to less than a quarter of what I sent to it from 9.10. And it ran just as well.

I think that those people had some other issues going on other than Docker overhead.

No. Corral was a complete rewrite of most of the system. Adding a few features to 9.10 isn't.

Well. It sounds like the current promise is that nearly all of Corral's great features can be plugged into 9.10 within a few months. Doesn't sound plausible to me. But, I would be happy to be proven wrong.

The whole release of Corral struck me as quite sudden.

How can that be? Corral was just the result of FreeNAS 10 which was available for testing for quite a long time.

but to knowingly release a "STABLE" NAS product without a GUI for disk replacement speaks to strikingly poor judgment on the part of whoever made the decision to release.

Kudos to iX for ultimately doing the right thing and killing it,

Why would that missing GUI element require killing the whole thing? Isn't it more sensible to just add the feature as a priority?

This is like saying that Grand Theft Auto should cancel the entire project if it was discovered that someone forgot to add an unpause button.

...and I'm probably speaking from a position of ignorance here. After all, I'm not a dev, I've never been a dev, and I'll almost certainly never be a dev. But I thought release practices in the software industry were relatively standardized.

No. I don't think there really is much standardization there. There are some common tendencies. But, individual projects/companies seem to have a lot of variation in their dev cycle strategies.

What are everyone else's thoughts about the UI?

It is very far behind where Corral is. It looks like a UI put together by developers instead of designers.

Keep in my new UI as far from finished. Its just a starting point (it was even based of an existing angular dashboard). A lot is still going to change (for the better).

I hope that it gets an investment into UX and graphic design...

 

[nas_scooter]

Love corral when it works good, but I've ran into several bugs that I'm sure other people have as well, and have spent a lot of time looking up or figuring out workarounds. I'm just as surprised as others that Corral is dead and that something new is in the works.

That being said, as a developer myself I know that things usually work out better when you take a buggy jumbled project and rebuild it to utilize more stable and complete frameworks.

I hope the new UI is as intuitive as Corral's, but with added stability and features. One of the major downsides to Corral's UI is the lack of mobile support. There have been a few times per week when I wanted to use my phone to log in to Corral to perform a particular function, but this was impossible with the Corral UI. I hope this is addressed in the new release.