Feedback about migration of 2 machines from ubuntu 24.04 lts to opensuse leap 16

Decided to try leap 16, so i switched two machines (one router, one server) from ubuntu 24.04. to leap 16.

What was a huge plus and what drove me to this was dropping of yast and focus on cockpit by the opensuse. While many opensuse users prefer yast way of configuration, the same has always been offputting and kept me off opensuse. From time to time i did use tumbleweed, but mostly i moved to fedora and ubuntu.

Migration was done 2w ago.

Keep in mind that this is a mix of personal opinions and subjective views, together with some comparisons (hopefully objective) with competition.

TLDR: opensuse moved into a good direction. will move more machines to it.

the bad

web

opensuse web is one of the worst things about opensuse. for example, compare the news part of the page between opensuse https://news.opensuse.org/ and a competitor Ubuntu Blog | Ubuntu.
this is not reflection on quality of suse software, but it is a very big repellent in today’s world where people expect good visuals. while this might be a subjective thing, i strongly believe it is not.

recent redesign of the landing page shows promise, but there are other issues:

  • wiki is sometimes terribly outdated (do not have the link to support this at this time), visuals seem rather weird as well as the links with colons
  • IDM login, this is really like in the government. once you create an account, changing anything is absolutely difficult because, for example, username is not kept in discourse (which is good, same for bugzilla), but idm itself has been poorely executed. finding a person that can help is… difficult.
  • download page seems like you had a bunch of postits with topics, thrown at the screen… what stuck it became the new page. visuals are really subpar here.
  • software part is about what? Search

functionality is not affected by the look and feel, but sure leaves opensuse in a disadvantage compared to competition when it comes to first impression.

conclusion: needs overhaul everywhere

no auto-update / restart in cockpit

opensuse leap dropped yast and embraces cockpit, but functionality which allows setting up of automatic updates and restart (like fedora), seems to be missing.

of course, many will recommend not to do this automatically, but this is super convenient for a stable distro that uses vanilla packages just with a few config files changed (samba, router, vpn…)

NetworkManager and systemd-network are in conflict

this was a rather difficult debug: i run a fedora box as a router, using systemd-networkd for configuring networking, dhcp server etc. at the same time, NetworkManager is enabled and doesnt interfere with anything. installing opensuse leap 16, i started to add same config files for systemd-networkd and started to lose connectivity to the machine. the reason was wrongly assigned routes by NetworkManager. when i deleted the connection with nmcli, this didnt help, so i disabled NetworkManager service and everything started to work fine. seems that it works in fedora, but doesnt work in opensuse.

second problem (?) is that now that NetworkManager has been disabled, network section in cockpit is disabled.

nmap mising in standard repos

zypper in nmap says nmap cannot be found

googling a bit, seems that it is not in standard repos. why?

firewall-cmd

firewall-cmd does not add devices to default zone. cannot confirm this, but devices were not added to any zone by default. im not sure if fedora does anything like this, but looks weird.

support duration / versioning

opensuse has some really niche offerings that i find really good: tumbleweed as a serious rolling release distribution. then you have something like rocky/alma/centos in the form of leap. great stuff.

so we have very fast upgrades and very slow upgrades. there is nothing inbetween.

every time some startup gets interviewed, ubuntu can be seen in the background monitors. what i understand, this 2y cycle, attention to desktop and having things user friendly are the reason why.

imho, ubuntu is doing the perfect balance for release cadence → lts is released every 2 years, which is important for 2 things:

  1. desktop needs to be kept more up to date
  2. cloud servers switch to newer tech more often since tech is changing fast. waiting for 5y to switch gives headaches with migration.

why no such desktop offering for opensuse?

a proposal: match ubuntu cadence and add leap as an extra migration path.

ubuntu naming also makes sense - i do not have to know when it is released and whatnot based on version like 16. there it is already given in the name… its simple, clear and to the point.

note: remember 42?

the good

support in bugzilla

had some interaction already via bugzilla for leap and i have to say response has been fast. kudos to people working in suse. a while back i also had some interaction with opensuse support and devs and it was always a pleasure.

one thing that is not so good is, when i assigned a ticket to security section, nobody touched it for almost 5months. only when i changed the classification away from security it got addressed. weird.

dropping yast

i said it once, going to say it again. dropping yast is a good thing. switching to systemd and cockpit is a good thing.

btrfs support in leap

fedora has it, rhel and centos do not. this is a very positive differentiation by opensuse. make this visible on your home page with benefits of btrfs.

would be nice to include encryption in btrfs without any sublayers (like dm-crypt) and unlock it with tpm on boot.

both, good and bad

X86_64-v2

moving to the baseline of X86_64-v2 seems a bit problematic since a lot of still usable hardware is now unsupported. probably enterprises do not need it since the long time ago move to better things, but in home environments this is still alive. it may have not been a decision made by opensuse but by suse, so you followed - if this is the case, try to communicate this and the reasoning. i dont find it affecting me personally, but i do feel sorry for people affected by this decision.

regardless of that, the line should be drawn at some point and leap 16 is a good time. there have been also discussions in fedora (i think) to do a similar thing.

what i find problematic here is that with this change you didnt drop grub2. if you have increased baseline, move the bootloader as well and support only systemd-boot.
grub2 was described by lennart as script that generates a script that generates a script. i got frustrated just by writing that.

move to selinux

this is both good and bad. good for security, bad for simplicity.

we need some gui tool to help with this.

not a leader

opensuse seems to mostly follow redhat in recent times. while i do not find this bad (i like redhat tooling), no innovation coming from opensuse camp is troubling. you should start investing into some new developments which are not just repackaging other things.

we can argue that redhat has NIH syndrome, but at least they did good decision for most of the tooling. new dnf5 is very good, i find it way more pleasant to use than zypper.

few thoughts opensuse can do to make news:

  • alternative dns caching server to dnsmasq
  • openwrt alternative built completely on opensuse and systemd (needs gui)
  • selinux gui tool
  • maybe start offering an alternative to postmarketos in packaging
  • learn from particleos (interesting talks in all-systems-go conference) and start implementing things that prove good
    • would work best as a tech demo / test for next leap in 2y if better release cadence would be selected
  • since you adopted a lot of redhat stuff, maybe think about adopting dnf5? (an idea, im also not sure if a good one)
  • restructure a bit main repo - since nmap is not in, i wonder what other basic tools need extra steps (btw, this pissed me off so much i considered to switch to fedora immedatelly, but then i had something to eat and moved on by using an other machine to do this - i still do not have nmap installed on that router).
  • btrfs encryption with tpm unlock. the thing bcachefs gave us hope, but then all hope was lost.

conclusion

i find opensuse moving into a good direction, so i will jump on-board with migrating a few more machines to it.

Zypper and Grub are two of the reasons I use primarily openSUSE. DNF and systemd-boot are two of the reasons why I do not use primarily Fedora. Like YUM, DNF is a mess of plugins and multiple man pages. Like with .deb systems, it’s whack-a-mole trying to determine what to do when something in package management outside routine on Fedora is needed. With zypper, few necessary plugins, so virtually everything is in one man page, trivial guessing where to look for instruction.

All installations here are multi- multiboot, no virtualization, so compatibility is seriously important. Legacy booting is done almost exclusively by Grub Legacy GFXboot in 100% Penguin mode from openSUSE’s retired releases. UEFI booting is done virtually exclusively through Tumbleweed’s Grub EFI, with sole /EFI/ on system containing usually only 2, sometimes 3 directories, and equivalent representation from efibootmgr. Bootloaders are not installed here on most installations other than Tumbleweed, and none are on currently supported Fedoras.

1 Like

@quantumflamingo

  • IDM/UCS with change at some point, it’s on the radar for the openSUSE Heroes Team…
  • Nope, read the Forum FAQ’s for details on any login issues…
  • Search is broken and will not be fixed, don’t use non-standard repos and all should be good and supported…

PS: You could get your css hat on and start pushing PR’s for web pages :wink:

  • nmap license changed, therfore it moved to the non-oss repository…
zypper se -s nmap
   | nmap                              | package | 7.95-lp160.2.1  | x86_64 | repo-non-oss (16.0)
  • For the Leap series, what versions are in Leap 16.0 today will in a lot of cases be around for the duration of the release cycle. Unless there is a critical CVE then there would be a version update to a package, else it’s just bug fixes, security fixes. It’s always been that way…
1 Like

somebody should usually means nobody is going to do it.

That’s how volunteer projects work.

That being said, while we (as developers) might take a look around and see what other projects are doing, and copy, borrow, and/or steal concepts and ideas from them, the argument “openSUSE should do this, because $other_project is doing it” is rarely a productive argument.

2 Likes

PS: You could get your css hat on and start pushing PR’s for web pages :wink:

I did something similar. I made an overall site concept using gohugo (I know opensuse uses different things, but I was studying gohugo and made it in that), made it public, opened a thread on forums, people told me to open issue on GitHub.

I opened an issue on GitHub, nobody said anything… that it’s good, that it’s bad, that it’s anything.

Since july

It’s one thing to open an issue and say “here’s an idea” - another to create the PRs to be evaluated and merged if they’re acceptable.

Starting from scratch is a huge project. Making changes to the existing design is lower effort, and as volunteers, everyone is busy not just with stuff related to the project, but with their day jobs.

2 Likes