While we were planning the upcoming UCS development stage, we decided to start working on the next major version: UCS 5.0 is planned for next year. In this article I would like to let you take a look behind the scenes and share some of our plans with you.
It’s been almost 5 years since we released UCS 4.0. During this time, UCS has evolved a lot. At the same time, we’ve continued to maintain the old version’s features. While most of them are popular with our users, others are not. There are also some things we would do differently if we had to do them again. By jumping to the next major version, we would like to get rid of some relics and implement several new features at the same time. We’re still at the very beginning, so not all decisions are final yet – but true to the motto “be open” I would like to share some of our ideas and plans in this blog post.
Changing to Debian 10 “Buster”
The new release will see an update of the base, and we will upgrade to Debian 10 (codename “Buster”), released in September 2019. In addition to all the upgraded packages, we will continue to reduce the differences between Debian and UCS in the main distribution. For example, Debian now has working UEFI Secure Boot (partly thanks to our support), so UCS no longer needs to make any adjustments.
Migration to Python 3
Most of our software at Univention is being developed in the Python scripting language. UCS 4.x uses the Python 2 runtime environment, although Python 3 has been around for quite some time. Python 3 is going to be the standard for implementations in UCS 5.0, so that we and our partners can benefit from the possibilities of the new version. You can see the first steps of this transition in UCS 4.4 as some of the packages have already been converted. Integrations or projects that make use of Python-based UCS interfaces such as UDM hooks or listener modules should run some checks until the release of UCS 5.0 to ensure Python 3 compatibility.
The UCS Online Demo gives you a quick and easy overview of the most important functions of Univention Corporate Server.
64-bit only
For some years now, we haven’t been offering 32-bit installation images for UCS (“i386 packages”), although we still offer updates for UCS 4.x. Introducing UCS 5.0, this support is coming to an end. To migrate 32-bit systems in your UCS Domain you should replace them one by one with new 64bit systems. Administrators should have completed the transition before upgrading to UCS 5.0. For all 64-bit environments, we will continue to support the upgrade of existing installations.
Bye-bye, unused Features!
Some of the UCS features are hardly used or not used at all. So we decided to take a closer look and only keep those features we definitely want to support in UCS 5.0. We’ve already made up our mind about those two features – they need to go:
- Access to EC2 instances via UVMM: UCS 4.x includes an extension for the UCS Virtual Machine Manager (UVMM) to manage instances in Amazon AWS-compliant IaaS environments. We will remove this integration.
- Support for NT-compatible Domains: UCS 4.x contains integrations, packages, and scripts for Windows NT domains in combination with Samba. We will not adopt these for UCS 5.
What is really used?
We will certainly remove more things from our distribution, because we believe we shouldn’t be spending our time maintaining unused features. So, can we please have your opinion and some answers to the following question:
Does a server system like UCS really need the KDE desktop environment?
Our product managers and developers appreciate your feedback. Thanks!
New Features
The new major release will not just see fundamental changes, but also contain innovative new features for the management system. Parts of these are going to be published for UCS 4.4 as well, stay tuned for upgraded apps and errata updates. We will report back in future blog posts.
Outlook and Roadmap
We’ve only just started working on UCS 5.0, so we can’t really announce a release date yet. One thing is for sure: it won’t happen before the next Univention Summit (January 23 and 24 2020, Bremen). Until then, we’re going to publish lots of exciting new features for UCS 4.4 and, of course, more announcements for UCS 5.0.
Comments
Christian Linen
Hopefully this doesnt cause our server to break, and is a smooth upgrade process in order to ensure continuity in the corporate office. Disturbances in server applications are not well received, so i do hope this is a priority so that we remain with your platform.
Ingo Steuwer
Hi Christian,
thanks for your comments.
Stable updates are a major objective for our development. The basic update process will be the same as for example for the update from UCS 4.3 to UCS 4.4.
As this is going to be a major release update, you might need more steps for preparation and postprocessing than in other upgrades – the release notes will describe what is needed. Please expect the update to take a bit longer as the amount of packages upgraded will be larger.
Hope that fits your needs!
Ingo Steuwer
Christian Linen
Corporate offices do not like big changes in order to add a new interface, for example. Once users learn a platform, it needs to stay relatively close or there is no ROI in here for us. it is wise to consider having to retrain users as a serious detrimet to the open source community. Users do not care about the extra stuff, only that what they are used to using is there, and does not radically change. Otherwise, it might be better to just host each app on its own copy of Debian to avoid your hassles.
Ingo Steuwer
Hi Christian,
thanks again for your feedback. To ensure we have the same understanding I’d like to explain a bit what users can expect from the update:
* 3rd party Apps hosted on UCS (Groupware, File Sync etc.; examples are Kopano, OpenXchange, Nextcloud, ownCloud, …): These Apps are going to be mostly stable even during an update of UCS itself. So while the underlying system gets many updates, the packages or docker container containing the App will most often stay the same or will se only minor upgrades to integrate with the new UCS version. This is the most prominent part for End Users – so they might not even notice an upgrade.
If Upgrades for 3rd party Apps are at the same time, we will try to split it to avoid situations where you are forced to upgrade both UCS and the App at the same time.
* UCS Core Services (examples are Active Directory compatible authentication, File, Print etc.): While these Apps fullfill an “everyday need” of many users, they don’t come with an dedicated interface. You might notice additional configuration options or new APIs, but our objective is always to try to stay compatible with the previous version. What will be improved is the web interface to manage these services, but as we don’t change the focus / use cases, I don’t expect prominent changes in the workflows.
* UCS Debian Distribution: The base packages will follow the normal Debian release upgrade process.
So, I fully understand your topic and we want to follow our process of continuous improvements in favor to “big disruptions” while improving UCS.
Beside that, I want to add that subscription owners of UCS 4 can expect a long term maintenance for UCS 4.4 so they will have enought time to prepare the update.
Ingo Steuwer
David Sugar
KDE is a curious question. Why would one want any desktop on a server? If I want a client experience, it will always be nicer on a desktop focused distro. If one is hosting thin clients, then I could see that, but that does not seem to me to be what ucs is about. I think it is better being focused on being a corporate server. Similarly, 32 bit is kinda dead too.
I like this roadmap in the broad sense. I also like that the value add is in python, and do appreciate the struggle with python3 migration. I have found myself in much the same situation with some of the things I am doing.
Ingo Steuwer
Thanks for your feedback!
In the past we had some feedback that a desktop environment even on a server is a good thing, for example to have a graphical tool to edit file ACLs or run a browser to open our management tools. But all feedback we got (most of it by mail to feedback@univention.de) was similar to yours – it’s not needed (anymore).
Alan Walters
Any sign of a date yet?
Christopher Etou
Hello Team,
Any release date for UCS 5.0?
Ingo Steuwer
Short update:
We’ve been focussed on upgrading the basis, beside the Debian Release upgrade we’ve spend a noteable amount of time on the Python 3 upgrade and related cleanups of our code base. Some of the work has been released as Errata for UCS 4.0. We also started to work on new features, including the Portal upgrade we released as Preview lately. And third we made some tough decisions about which features to improve and which to discontinue. Based on this we’ve narrowed down the planning and will release UCS 5.0 in Q4/2020. I’ll write more about it in a separate blog post in the next weeks!
Ingo Steuwer
Update on our Release plan: https://www.univention.com/blog-en/2020/10/ucs-5-0-status-update-beta-release-in-december-2020/
lonnie ray
This upgrade has caused severe upheaval in our medium office. We still cannot get owncloud running properly again, and is now basically useless with the default 2mb upload limit. After 3 weeks of battling UCS in full view of the corporation, not sure we will stick with it. There is nobody in the help forum answering questions anymore. We cannot even get the upload limit reset so it is even usable. Looking for another answer now.