Understanding Proxmox 3.4 EOL and 4.0

[adamb]

My management would like to better understand what EOL means for Proxmox 3.4? There isn't a ton of details out there. I am pretty sure its April 2016, but what does this entail? Does this mean 3.4 clusters will no longer be supported by proxmox? Will we not get any security related updates? Just trying to get a better understanding.

Am I also correct in thinking that Proxmox 4 no longer is based on the rhel kernel at all?

They are also questioning the reasoning behind making such a huge jump to a new kernel, any input on that? We are a CentOS/Rhel shop, so big jump in kernel version scare the heck out of management.

 

[mir]

If EOL for 3.x follows the same path as for 2.x then EOL means no more security or bug fixes by that date.

The kernel for proxmox 4 is based on the latest Ubuntu server LTS kernel.

 

[adamb]

If EOL for 3.x follows the same path as for 2.x then EOL means no more security or bug fixes by that date.

The kernel for proxmox 4 is based on the latest Ubuntu server LTS kernel.

Wow that is quite disappointing. Hoping the developers can provide exact details.

 

[mir]

What is disappointing, the kernel or the EOL?

 

[adamb]

What is disappointing, the kernel or the EOL?

The EOL. We have 3.4 clusters out in the field that aren't even 1 year old yet (3.4 was out 2/15 of this year), puts us in a terrible position and is making us question proxmox in a enterprise environment.

We are also looking at deploying a very large project based on proxmox and we are running into issues with 4.0 that the Dev's are digging into (which I have confidence they will pin it down). This is giving my management cold feet on 4.0, they want to stick with 3.4, but that doesn't look to be an option.

 

[mir]

I can easily recognize you point but the EOL date is closely connected with EOL for Debian Wheezy which proxmox 3.x is based on. But given the fact that Debian Wheezy will continue as an Debian lts until May 2018 maybe the devs will reconsider. Read more here: https://wiki.debian.org/LTS

 

[adamb]

I can easily recognize you point but the EOL date is closely connected with EOL for Debian Wheezy which proxmox 3.x is based on. But given the fact that Debian Wheezy will continue as an Debian lts until May 2018 maybe the devs will reconsider. Read more here: https://wiki.debian.org/LTS

Ahhh see I was thinking Debian was 5 years without LTS. My mistake. Still, hard to understand how anyone can accept this as reasonable in a enterprise environment. My management is going to have a hay day with this, bummer.

 

[dietmar]

My mistake. Still, hard to understand how anyone can accept this as reasonable in a enterprise environment. My management is going to have a hay day with this, bummer.

It is quite easy to update from 3.X to 4.X. This may involve some work on your side, but in the end you get better software with more features...

 

[adamb]

It is quite easy to update from 3.X to 4.X. This may involve some work on your side, but in the end you get better software with more features...

We currently have 17 installs out in the field, its really not that easy. Inhouse stuff isn't that big of deal.

Are security updates still handled by LTS?

 

[tom]

Ahhh see I was thinking Debian was 5 years without LTS. My mistake. Still, hard to understand how anyone can accept this as reasonable in a enterprise environment. My management is going to have a hay day with this, bummer.

Proxmox VE is following a rolling release update model. Only the current version is supported. Due to the replacement of OpenVZ and the new HA manager, we decided to support also 3.x for some time. And yes, currently the plan is to stop this on 4/2016. Please note, Debian Wheezy support will also end on this date. Debian Wheezy LTS will be available, but this is only for "some" packages, never for the full distribution (to my best knowledge).

In contrast to other enterprise distributions, we always provide a fully supported upgrade path. The upgrade from 3.x to 4.x is not as smooth and easy as we would like but its still possible without re-installation - our team got good experience here to assist.

In the case of discussion for your management: Other so called "enterprise distribution" cannot provide such supported in-place upgrade paths. e.g. some (or all?) Windows Servers, also RHEL servers are not that famous for the in-place upgrade paths (in-place upgrade it partially possible from RHEL6 to RHEL7, but only in some cases and more or less never in HA setups.)

Both models exists in the industry, but we think our model is better suitable for our users, so therefore we went this way since the beginning of Proxmox VE in 2008.

 

[adamb]

Proxmox VE is following a rolling release update model. Only the current version is supported. Due to the replacement of OpenVZ and the new HA manager, we decided to support also 3.x for some time. And yes, currently the plan is to stop this on 4/2016. Please note, Debian Wheezy support will also end on this date. Debian Wheezy LTS will be available, but this is only for "some" packages, never for the full distribution (to my best knowledge).

In contrast to other enterprise distributions, we always provide a fully supported upgrade path. The upgrade from 3.x to 4.x is not as smooth and easy as we would like but its still possible without re-installation - our team got good experience here to assist.

In the case of discussion for your management: Other so called "enterprise distribution" cannot provide such supported in-place upgrade paths. e.g. some (or all?) Windows Servers, also RHEL servers are not that famous for the in-place upgrade paths (in-place upgrade it partially possible from RHEL6 to RHEL7, but only in some cases and more or less never in HA setups.)

Both models exists in the industry, but we think our model is better suitable for our users, so therefore we went this way since the beginning of Proxmox VE in 2008.

Thank you for providing some input I can actual go at management with.

 

[dietmar]

We currently have 17 installs out in the field, its really not that easy. Inhouse stuff isn't that big of deal.

Well, should be worth the effort (considering the minimal costs from our side ()).

Are security updates still handled by LTS?

Debian LTS will provide security updates for basic system. We will provide kernel/kvm updates for some time, but that is not finally decided.

 

[adamb]

Well, should be worth the effort (considering the minimal costs from our side ()).




Debian LTS will provide security updates for basic system. We will provide kernel/kvm updates for some time, but that is not finally decided.

Has nothing to do with effort or cost. I agree you guys provide a pretty solid upgrade path, im hoping that is enough to sway management. They are very use to a 10 year life span due to rhel/centos.

 

[mir]

We also have several hundreds RHEL servers at my work ranging from EOL 4 (due to some oracle portal and forms stuff) to 5 and 6. But notice that an upgrade from one major RHEL to another is neither supported nor certified. So yes, you get 10 years of support but no upgrade path. For Windows server the support period isn't even official which means they can remove support over night, and again, no supported or certified upgrade path.

 

[adamb]

We also have several hundreds RHEL servers at my work ranging from EOL 4 (due to some oracle portal and forms stuff) to 5 and 6. But notice that an upgrade from one major RHEL to another is neither supported nor certified. So yes, you get 10 years of support but no upgrade path. For Windows server the support period isn't even official which means they can remove support over night, and again, no supported or certified upgrade path.

Yea its definitely a good point, it makes me happy.

 

[TimSmall]

Debian LTS will provide security updates for basic system. We will provide kernel/kvm updates for some time, but that is not finally decided.

Hi,

I was wondering if there was any update on this? Wheezy is now entering LTS...

Wheezy LTS EOL: May 2018
OpenVZ RHEL6 kernel EOL: Nov 2019

(sources: https://wiki.debian.org/LTS and https://wiki.openvz.org/Download/kernel/rhel6 )

I assume you're using the OpenVZ RHEL6 kernel for Proxmox 3.x ?

I'm currently managing a few OpenVZ hardware nodes for a client (with in-house pacemaker + drbd configuration - which we've been maintaining for >5 years now), and we are considering moving to Proxmox. We've made the odd minor contribution to pacemaker, and OpenVZ in the past, so I'm hoping we'll be able to do the same with Proxmox.

Whilst I do have some LXC experience, I'm assuming a smoother path for us will be do go with 3.x, as that is a lot closer to the current stack we're running, but it would be good to get some clarity about your plans in connection with Wheezy LTS? Obviously if it's 2 months, followed by no support, this isn't going to be viable!

It would be good to get some clarity on what will be supported (by way of your custom features) for security updates, and over what time span? e.g. no ongoing zfs or kvm support would be fine for us (as I understand that neither of these are being pulled from your upstreams).

Cheers,

Tim.

 

[Q-wulf]

I was wondering if there was any update on this? Wheezy is now entering LTS...

Latest discussions can be found ..

• Here https://forum.proxmox.com/threads/moving-to-lxc-is-a-mistake.25603/
• And here https://forum.proxmox.com/threads/when-will-3-x-become-eol.25852/

 

[mo_]

We also have several hundreds RHEL servers at my work ranging from EOL 4 (due to some oracle portal and forms stuff) to 5 and 6. But notice that an upgrade from one major RHEL to another is neither supported nor certified. So yes, you get 10 years of support but no upgrade path. For Windows server the support period isn't even official which means they can remove support over night, and again, no supported or certified upgrade path.

While RHEL has this no-upgrade-thing going on, SLES certainly allows upgrades. You can upgrade SLES 9 to 10 to 11.4 just fine (SLES 12 isnt recommendable) and you still get quite a bit of support for those.

The problem is that IT architecture tends to be rather slow moving or static alltogether. The bigger the company, the slower it gets due to ITIL processes and other forms of change management. For instance: a current client of mine is only just now *planning* to replace/upgrade SLES9-VMs (SLES 10 LTS support will end July this year...)... For what its worth, theres still a small number of AIX 5 systems too

 

[e100]

Debian LTS will provide security updates for basic system. We will provide kernel/kvm updates for some time, but that is not finally decided.

I think you should provide kernel/kvm updates during wheezy LTS and set a solid date, at least one year in the future, when you will discontinue providing those updates. Ideally that date would coincide with wheezy LTS EOL.

Currently, LXC in Proxmox 4.0 does not provide the level of isolation that OpenVZ did, that makes it unsuitable for untrusted guests.
DRBD 9 is too new, some of us do not feel comfortable using it in Production yet, Even Linbit says its not yet stable, http://drbd.linbit.com/home/roadmap/

 

[CTCcloud]

Proxmox VE is following a rolling release update model. Only the current version is supported. Due to the replacement of OpenVZ and the new HA manager, we decided to support also 3.x for some time. And yes, currently the plan is to stop this on 4/2016. Please note, Debian Wheezy support will also end on this date. Debian Wheezy LTS will be available, but this is only for "some" packages, never for the full distribution (to my best knowledge).

In contrast to other enterprise distributions, we always provide a fully supported upgrade path. The upgrade from 3.x to 4.x is not as smooth and easy as we would like but its still possible without re-installation - our team got good experience here to assist.

In the case of discussion for your management: Other so called "enterprise distribution" cannot provide such supported in-place upgrade paths. e.g. some (or all?) Windows Servers, also RHEL servers are not that famous for the in-place upgrade paths (in-place upgrade it partially possible from RHEL6 to RHEL7, but only in some cases and more or less never in HA setups.)

Both models exists in the industry, but we think our model is better suitable for our users, so therefore we went this way since the beginning of Proxmox VE in 2008.

Tom hits the nail on the head here so to speak.

The biggest reason that other corporations have to support for 5 years or more is exactly as he mentions...no clean upgrade path which means it's a migration which means major planning. Being that Proxmox has such an AWESOME track record of stability and has a clear upgrade path, that in and of itself is better than supporting a version for 5 years because the upgrade is minor in comparison to a migration to a new version.

We've successfully upgraded from version 2.x to 3.x and are now on 4.x and I didn't a have any issues each time and we've got various hosts in the field as well and although it had to be planned for each case, it was still dead simple in comparison to a migration.