11.2-RELEASE-p9 End-of-Life

noodlefling

Active Member

Reaction score: 11
Messages: 145

Code:
WARNING: FreeBSD 11.2-RELEASE-p9 is approaching its End-of-Life date.
It is strongly recommended that you upgrade to a newer
release within the next 2 months.

I got the above warning message when doing a freebsd-update fetch (although no updates were needed) today.

The site says 11.2 will be good until "11.3-RELEASE + 3 months", but 11.3 isn't even out yet.

Should I just ignore the message for now? Major version upgrades often come with unnecessary pain (I never have any interest in new features, I just don't want the system to be insecure), so I'd like to hang on to 11.x for as long as possible while the bleeding edgers beta test 12.x.
 

Vull

Aspiring Daemon

Reaction score: 434
Messages: 730

According to its release schedule 11.3-RELEASE won't be released until some as-yet-to-be-announced date on or after July 9, so we should still have until some time on or after October 9.
 

kfv

New Member

Reaction score: 5
Messages: 6

According to the FreeBSD support model, yes 11.2-RELEASE will be supported at least three months after 11.3-RELEASE so you have nothing to worry about and it's only an announcement since 11.3-RELEASE itself would not be out within two months.
To stay tuned on the release process follow the release schedule.

In the meanwhile, 12 is not considered beta and personally am using it in production environment, so give it a try if you like ;-)
 
OP
noodlefling

noodlefling

Active Member

Reaction score: 11
Messages: 145

I guess it's a suggestion and not a requirement, but it's a weird error message.

Anyway, thanks for the confirmation, guys.
 

Chris236

Member

Reaction score: 9
Messages: 32

It shows the utter absurdity of the FreeBSD support model.

Somebody needs to look at the models of CentOS or Ubuntu.... we need support frames of 5-10 years for professional environments.
 

PMc

Daemon

Reaction score: 676
Messages: 1,357

It shows the utter absurdity of the FreeBSD support model.

There should be no "support" model. If somebody needs support, then in any and all realms of life it is the case that they have to engage somebody to provide them with their needed support. TANSTAAFL.
 

Chris236

Member

Reaction score: 9
Messages: 32

There should be no "support" model. If somebody needs support, then in any and all realms of life it is the case that they have to engage somebody to provide them with their needed support. TANSTAAFL.
Sorry that is nonsense. The second the FreeBSD project declares a release EOL, I can not buy support for it any more. Believe me, I have tried. The lack of affordable commercial support for FreeBSD is staggering, and the very few vendors who can be persuaded to provide it (usually,a t much inflated prices) always categorically demand to cover only non-EOL releases.

Oh - and I am not talking "support" support. I am talking security patches support.

And I would LOVE to throw the kind of money we throw at Redhat towards a BSD support organization - but that just does not exist, not in the US and certainly not in Europe. The vendor catalog that used to be on the FreeBSD site (is it still there?) was always an outdated joke with most companies on it having no clue about them being there, and the rest being small consultancies or freelancers without the substance for serious support contracts.
(And again, believe me, I tried phoning all the Central European ones at some point in the past)

And given all that, your hastily thrown down TANSTAAFL comment is outright offensive. You have no clue what you are talking about. I pay for my lunch, thank you. There is just no vendor to buy it from.
 

PMc

Daemon

Reaction score: 676
Messages: 1,357

Sorry that is nonsense. The second the FreeBSD project declares a release EOL, I can not buy support for it any more. Believe me, I have tried. The lack of affordable commercial support for FreeBSD is staggering, and the very few vendors who can be persuaded to provide it (usually,a t much inflated prices) always categorically demand to cover only non-EOL releases.

You're welcome - I know that it is nonsense. But it is a problem, and I don't really see a solution: anywhere in life when somebody needs some kind of support, they pay for that - and in most cases they pay willingliy. But how could such a scheme be implemented into FreeBSD?

Oh - and I am not talking "support" support. I am talking security patches support.

That makes things more complicated, but otherwise it's the same: there is the need to put deliberate effort into some work, and there are -probably- people who would pay for a proper outcome. But while we might manage to bring them together for the matter of individual support, with security patches it is a collective matter.

And given all that, your hastily thrown down TANSTAAFL comment is outright offensive. You have no clue what you are talking about. I pay for my lunch, thank you. There is just no vendor to buy it from.

Okay, sorry for that. I never looked at it from that viewpoint, I only noticed that there is no market for *BSD. So either the userbase is too small - or people just take for free what they get for free (those who copy the code do so, that's obvious), or - I don't know. From legal terms (as I think I understand them) anybody would be free to do with *BSD the same thing as RedHat did with Linux.
 

SirDice

Administrator
Staff member
Administrator
Moderator

Reaction score: 12,270
Messages: 38,773

we need support frames of 5-10 years for professional environments.
5 year support on the base OS is actually irrelevant. The ports tree and everything it contains doesn't fall in this category. You have the exact same applications regardless of the version of the base OS. And those are the ones that are important, not the version of the OS. I mean what good does 10 year support on FreeBSD X.Y do when you're depending on PHP, MySQL or any of the other applications which don't follow that support model?

Does it really matter if you have FreeBSD 11.2 with Apache 2.4 and PHP 7.1, FreeBSD 11.3 with Apache 2.4 and PHP 7.1 or FreeBSD 16.9 with Apache 2.4 and PHP 7.1?
 

PMc

Daemon

Reaction score: 676
Messages: 1,357

The fact itself does usually not matter, but upgrading means that you need to do extensive tests concerning, what does the new release do with the hardware.
The software packages are not so big a problem, they are data-in/data-out, and if in doubt one can look at that data. But the device drivers are data-in/electricity-out, and it is a bit more difficult to test that.
 

tommiie

Well-Known Member

Reaction score: 85
Messages: 297

Major version upgrades often come with unnecessary pain

I'm quite new to the party here at FreeBSD. I installed 11.2-RELEASE on my laptop in January 2019 and was quite reluctant to upgrade to 12.0-RELEASE, afraid I would break all kinds of things. The upgrade, however, went very smoothly. Except... that it broke my manually compiled i915kms driver which I needed as the base driver from 11.2 was too old for my hardware. That was easily fixed though by installing the base driver from 12.0.

I've also been told it would be easy to "downgrade" should you have ZFS installed. Something with snapshots and such. I haven't tried this yet but it sounds very cool. So perhaps you could give it a shot anyway on a few of your servers?
 
H

hukadan

Guest


Well, IBM is in the business of making money, not giving it away. So long term support must be a good idea., isn't it?
As long as you have enough money to hire people to make it happen. Such long time support does not happen in vacuum.
 
  • Like
Reactions: kfv

ralphbsz

Son of Beastie

Reaction score: 2,337
Messages: 3,233

Let's ignore the whining about support models, and instead focus on the OP: That message is partially misleading, and partially outright wrong. I saw the same message Friday night on my machine. Let's analyze it:

"FreeBSD 11.2-RELEASE-p9 is approaching its End-of-Life date."
That's correct, but only if you use a very aggressive definition of end-of-life. That date is July 5, 2019 + 3 months, or a little over 6 months away. Given that the grace period for minor releases is 3 months, the term "approaching" is a misleading exaggeration. But not wrong in the mathematical sense: October 2019 is indeed approaching, as is January 19, 2038, the time in a few billion years when our sun will become a red giant, and the heat death of the universe. Showing this message in August or September would be reasonable, today it is not.

"It is strongly recommended that you upgrade to a newer release within the next 2 months."
That is simply wrong. Within the next two months, there will be no newer 11.X release; it only shows up in 2 months + a few days.

Someone who really cares should file a PR against that message, or volunteer to get it fixed. I don't happen to care, since I know how to read the release schedules.
 
OP
noodlefling

noodlefling

Active Member

Reaction score: 11
Messages: 145

I appreciate that they're continuing to keep the OS active and viable, but it does seem like we are running through major versions a lot faster than we used to.

Maybe I'm just getting older and time is dilating for me, but it seems to be more of a nuisance than it used to be.
 
  • Thanks
Reactions: OJ

sidetone

Daemon

Reaction score: 906
Messages: 1,855

I have a concern now.

I need to reinstall FreeBSD. If I install 11.2, it will not be supported in a few months. I downloaded 12.0 for the cd, and it doesn't fit on one cd. I'll have to go with the boot cd, but those installs usually take longer, bc it needs to download as it goes.

When I used to install OS'es from flash drives, the flash drives usually got data or data corruptions written on them during bootups if it crashed and I had to reboot, or if for another reason. I know this is not a problem for harddrives, but it is not common to install OS'es directly from them like this. The cd is read only, so the data doesn't change no matter what. For whatever reason, the flashdrive with the install needs to be reliable indefinitely.

If 12.0 has pkgbase, the cd medium shouldn't be that big.

The data being too much for the CD medium it is meant for is usually a problem for stable releases, prereleases, on DragonFlyBSD, and in this case for FreeBSD 12.

It seems like pkgbase can be used in the future for one continueless and maybe versionless quick rollout.
 

recluce

Active Member

Reaction score: 40
Messages: 170

It shows the utter absurdity of the FreeBSD support model.

Somebody needs to look at the models of CentOS or Ubuntu.... we need support frames of 5-10 years for professional environments.

Ubuntu, really?

True, every 2 years there is a LTS version that receives some support for five years. Many packages are not receiving support nearly that long. In any case, you pay by having a significantly outdated OS. Once the five years are over (or if you decide to upgrade earlier, due to outdated software), the version upgrade usually breaks things to a point that a fresh install is inevitable or at least a really good idea. I haven't had a single version upgrade of Ubuntu and its flavours that worked as expected.

By comparison, unless following -CURRENT, I have never had issues with any minor version update of FreeBSD. Even with major version updates, I can't think of any serious problem. One of my servers has now gone through every version from 9.3 to 11.2 (except 10.0 and 11.0) and no issues. On a more experimental desktop, I went from 10.1 through 11.x and 12-CURRENT to 12-STABLE. CURRENT caused issues with updates, as expected. Rock stable again since 12-STABLE.

So comparing Linux updates / upgrades with FreeBSD updates is comparing apples and oranges.
 
  • Like
Reactions: kfv

ralphbsz

Son of Beastie

Reaction score: 2,337
Messages: 3,233

I need to reinstall FreeBSD. If I install 11.2, it will not be supported in a few months.
You could just install 11.2 right now, and then upgrade it in place to 12.X if you want.
When I used to install OS'es from flash drives, the flash drives usually got data or data corruptions written on them during bootups if it crashed and I had to reboot, or if for another reason.
Strange. While I've had USB or SD storage fail, it tends to be complete hardware failure (they become dead), not data corruption. I see no logical reason why a USB drive that's mounted readonly should be corrupted during a boot up.

I know this is not a problem for harddrives, but it is not common to install OS'es directly from them like this.
Actually, why not? Buy a cheap external hard drives (I regularly see ~TB drives at Costco in the US for $60), and use it as your install media. Would that work?
 

neel

Active Member
Developer

Reaction score: 57
Messages: 135

Ubuntu, really?

True, every 2 years there is a LTS version that receives some support for five years. Many packages are not receiving support nearly that long. In any case, you pay by having a significantly outdated OS. Once the five years are over (or if you decide to upgrade earlier, due to outdated software), the version upgrade usually breaks things to a point that a fresh install is inevitable or at least a really good idea. I haven't had a single version upgrade of Ubuntu and its flavours that worked as expected.

By comparison, unless following -CURRENT, I have never had issues with any minor version update of FreeBSD. Even with major version updates, I can't think of any serious problem. One of my servers has now gone through every version from 9.3 to 11.2 (except 10.0 and 11.0) and no issues. On a more experimental desktop, I went from 10.1 through 11.x and 12-CURRENT to 12-STABLE. CURRENT caused issues with updates, as expected. Rock stable again since 12-STABLE.

So comparing Linux updates / upgrades with FreeBSD updates is comparing apples and oranges.

On top of that, I found FreeBSD (including CURRENT) to be more stable than Ubuntu LTS. I had my fair share of CURRENT breaking, but nowhere near the issues with Ubuntu. I had Ubuntu 16.04 on my "school" laptop (that I use for college) until it was unable to boot (the laptop ran Fedora for two months and now runs FreeBSD 11.2 but I probably should upgrade it to 12.0)..
 

kfv

New Member

Reaction score: 5
Messages: 6

I need to reinstall FreeBSD. If I install 11.2, it will not be supported in a few months.
As ralphbsz mentioned, you can upgrade to 12.X whenever you please. But please note 11.2 is a minor release of 11 branch, meaning that you could stay on 11 as long as the branch is supported - which is 2021 for stable/11.
All you need to do is upgrading to newer minor releases, for instance to 11.3 when it's out, so not a big deal mate, you're safe on 11.

It shows the utter absurdity of the FreeBSD support model.
It's called policy, not absurdity.

Somebody needs to look at the models of CentOS or Ubuntu
Negative. You're comparing totally different things.

we need support frames of 5-10 years for professional environments.
Five to ten years for the base OS? Could you please give us an example when one would need a ten years of support? I think it's better to invest on more important projects and I believe FreeBSD Foundation is doing a great job.
Do you know what the cost of supporting a base OS for a decade will be? Although I disagree with you on needing a support model like linux distros, I understand your concern, but as I said earlier, you're comparing totally different things and there's a lot to consider.
 

sidetone

Daemon

Reaction score: 906
Messages: 1,855

Actually, why not? Buy a cheap external hard drives (I regularly see ~TB drives at Costco in the US for $60), and use it as your install media. Would that work?
Too much hassle. To put an iso image on a harddisk, that has more space than needed, then to get the bios to recognize the drive how the data wants to be recognized.

I'll just use the boot cd install.
 

Chris236

Member

Reaction score: 9
Messages: 32

Five to ten years for the base OS? Could you please give us an example when one would need a ten years of support?
Lets introduce some reality. All numbers below are optimistic.
  • 0 months: a new FreeBSD release is published.
    to use it I need a type acceptance and engineering -> budget
    budget is requested 6-12 months in advance.
  • +6 months: budget for testing and release engineering available for odering
  • +4 months: external consultant is purchased and onboarded for project
  • +3 months consultant has designed and aligned test specification,
    engineered an image prototype, created a test environment
  • +3 months consultant has completed first full test run, documented results, fine tuned system, implemented hardening according to company guidelines, implemented and tested management integration in test field.
  • +1 month Acceptance tests. I get type acceptance and release documentation and image.
  • +0 at this point in time, application project that actually needs the OS falls fully engineered and ready for deployment from heavens. It also ran its engineering, tests and type acceptance concurrently to the OS and there was zero interdepenance time cost for both projects running in parallel. (Miracle #1)
  • +0 also at this point, 50 servers are deployed in 10 unmanned tech sites all across the country, every single one fully and error-free connected to the network and with suitable remote control hardware configured and connected. (Miracle #2)
  • +1 month all 50 servers have been installed and successfully brought online with OS and loaded with various parts of application software. Management, alarming, monitoring, and logging is configured and integrated in central systems. Provisioning is activated and tested.
  • +3 months application engineers integrate the app, make everything work together, do test runs, find and fix issues not found before
  • +1 month field acceptance: operations takes over, does backup and recovery tests, bare metal restore, component failover tests, full system failover tests, does final hardening, finalizes operational documentation. Declares ready for service.
  • +1 month Marketing determines launch date, announces new functionality etc. System is launched in production.
All those numbers are optimistic. Some very much so. Still, we have the first server in production about 2 years after the release was announced. Reality is sometimes it takes longer. Now we need that stuff to just work and earn money. The 50 servers are run by a 3-man Operations team that also runs another 5-10 or so other such applications. Typical life cycle for such a piece is 5-7 years, 10 - 15 is not unheard of.

Nobody has time for frequent updates here. If any, updates are centrally scripted and deployed at substantial cost. freebsd-update is a pretty bad customer there, as it needs serious amounts of manual intervention for a full upgrade. The reference is is Redhat Sattelite (or maybe even WSUS)

So, yes! We need 10 years support, at least for required security patches and a functioning ports and packages framework. And, as the Redhat example has shown, people are willing to pay for it.
 

shkhln

Daemon

Reaction score: 964
Messages: 2,217

+3 months consultant has designed and aligned test specification,

I understand an OS release can't be tested before it has been released (obviously), but why would someone postpone designing tests for that? What these tests actually do?
 
Top