My experience with Tumbleweed, so far

Firstly, a confession: I tried unsuccessfully to install openSUSE over 5 years ago and kept getting vague permanent errors as I recollect: Cannot continue installation due to permanent errors. It only gave me one option, Abort Installation. So of course, openSUSE remained uninstalled for half a decade. Having some time with Arch I swore off rolling releases until I started to read up on openSUSE’s history and Tumbleweed. Could this be what I have always been looking for all along? Inbuilt and preconfigured rollback from the GRUB menu. Ok, Silverblue does that, but it’s an immutable core and does not play well with file permissions.

So this snapshotting utility called Snapper was developed by SUSE which could have saved my Arch installation the day of the dreaded GRUB boot failure in 2023 where even their help pages could not get me back into arch-chroot, so my data losses were serious as I was at the time working on a very intense dynamic project that was changing by the hour. I never want to be in this situation again.

Fast forward today, I just installed Tumbleweed a few days ago. It was fast to install, with the most comprehensive setup page I have ever seen, I was able to choose everything I wanted with granular precision. Then I opened YaST and I was floored by how many settings can be managed. It did not manage to detect my network printer with nothing found, same with scanner. A tutorial on how to open the appropriate ports in firewalld would have been helpful, but I am sure it exists and I haven’t worked out how to find it yet. Still I managed to work it out with the excellent documentation and got the printer working with IPP-Everywhere by simply entering the IP address. The scanner was a little more tricky until I found out about sane-airscan from the Discord channel. So it’s all good.

Now to make my system completely impermeable against disaster apart from snapper. What about openSUSE Rescue System? Is there a Tumbleweed version of it or is it just SLES? or what about Clonezilla for everything? What about Borg Backup as I hear it does not have the limitations of reading BtrFS like Timeshift has. I’d love to get your ideas on it and of course, comment on my ramblings and how to best prepare for that inevitable day of disaster! I’m just an old retired geezer who loves technology.

Michael

2 Likes

I know many will disagree, but Tumbleweed is the wrong OS for that! If you want ROCK SOLID, use Debian or Leap.

I run Leap. For my personal needs, I find btrfs snapshots to be very cool, but useless. I do imagine snapshots to be useful for Tumbleweed, as sooner or later, you will get an update that breaks something you actually need. Booting into a pre-update snapshot is a nice and fast way to postpone figuring out what actually went wrong, or keep you working while waiting for a fix (most likely an updated kernel module).

As far as system back-up goes, I am rather inconsistent. Sometime I use rsync; sometimes I use gnome-disks to create a images of “/”, “/boot”, and “/boot/efi”. I have used timeshift in the past and was pleased with it, although I never had to use it (or anything else) for recovery. It can be configured to use rsync instead of btrfs snapshots.

For my non-system personal data, as in “/home” and other filesystems, I just use rsync to external drives. I am seriously considering ZFS duplication as an alternative, provided I can get ZFS to work on Leap … gotta love new geeky projects like that :smile:

Cheers

I have had zero issues with Tumbleweed here, I don’t use snapper, just btrfs and xfs (for /home and a couple of other locations).

Maybe a couple of minor issues at times when running X11 (On wayland now) and forgetting to change PCIe id’s in xorg.conf, but nothing stopping the system boot to at least multi-user.target.

I do use rsync over ssh for data backup to another system, not worried about the OS if it goes pear shaped and it hasn’t since it was installed 3 odd years ago.

Look at btrfs send…

1 Like

Agreed that Leap is more solid than Tumbleweed but unless you have a Nvidia graphics card I personally think Tumbleweed is very solid. The only think that sometimes breaks are lesser used packages and more often Nvidia support.

Apart from snapper I hope you did create a separate home partition. With that a possible re-install is very painless, did do it multiple times but almost could not noticed I had reinstalled, the desktop did not change. Only thing you have to do after reinstall is install the packages you manually installed and the settings you made.

1 Like

@marel Can’t beat that run file install :wink: works every time, no waiting around for rpms and I use the open version, albeit for prime Render Offload…

1 Like

We used to have TW on all our machines.

We switched all our machines to Leap 15.6 and will no doubt stay there. The constant requirement of “zypper dups” (including the vast amount of sofware packages being updated) and battling the occassional issue, just doesn’t appeal to us anymore.
(if you don’t believe me, just observe all the “problem” threads in the TW forum for a couple of weeks, compared to the Leap forum :slight_smile: )

The fancy GUI installer (and its detailed options), plus the use of Yast2 (after installation) is NOT specific to TW - it’s the same with Leap/etc.

As far as “system rescues” go … we use XFS for our /home filesystem (yes, /home is a separate partition), and everything else is BTRFS.

We keep four bootable kernels at the ready, selectable at GRUB. (we learned to do that while using TW).

To continue the “system rescue” thing, we only backup the /home partition, and we use rsync twice a week to backup /home. After the initial rsync backup, the subsequent backups go fast.

If something catastrophic happens with the root filesystem, and can’t go back to a snapshot, we would simply do a fresh install.

TW is great if you enjoy a rolling distro, like to be at the cutting edge, and don’t mind running zypper dup all the time. If you go on vacation for a couple of weeks+, be prepared to wait on a huge update to finish :slight_smile:

5 Likes

No, I just followed the default install apart from my patterns. I just save /home/$USER to external storage. I will install Borg Backup to handle that in the long term. What do you think?

A separate (non-btrfs) home partition is I think better but without it it will also work fine.

For btrfs systems, make sure you have enough free space, you do not want to run into low disk space with btrfs, that is something btrfs does not handle nicely.

If you have experience with Borg Backup (incl. doing a restore), go for it, if not I would suggest something more basic like rsync.

1 Like

Hello @aggie ,

It is however, unique to suse/opensuse. It absolutely amazes me that opensuse continues to be the ONLY gnu/linux project with a genuinely good installer.

While I think YaST as a general config/management tool is in serious need of some TLC, I am not aware of another tool that even comes close.

YaST and zypper are the primary reasons I prefer Leap over Debian. The relatively easy firewall configuration comes in as a close second.

Cheers!

1 Like

There are also other linux distributions which provides modular control centers and installers. As example Mageia (former Mandrake/Mandriva) with their „Mageia control center MCC). Also PCLinuxOS uses an adapted variant of it. openSUSE is not the only distribution with such features.

Hello @hui,

I’ll have to look into Mageia at some point. A quick glance through the install docs … it looks decent. My claim of “the only” is much more about installation. In my admittedly limited experience, other installers lack granularity and fall horribly short with anything other than installing on plain disks, such as “root” on raid, etc.

Hear, hear, I hope the devs do as good a job with the new Agama installer which is in the works. It’s hard to beat the current installer’s filesystem/partitioning features and layout. Rock solid and very easy to use while being advanced enough not to hinder any unusual setups. :rocket:

Another older guy, no computer education since the “punch card” days back in '73 . . . at that time, my adolescent insight was, “This isn’t going to get us to Star Trek levels of computer technology anytime soon.” : - ))

Been kicking around in linux since maybe '07 . . . worked my way up from the ubuntu PPC days to now 7 distros done on a machine multi-boot style . . . a TW install handles my grub management, with a Leap, there was a Slow Roll until it blew itself up, a Debian Sid and a Trixie, a Manjaro, a Mint LMDE, and a Lubuntu Oriole . . . on three drives.

I agree with those who say “TW isn’t ‘stable’ Leap is” . . . but then, on the relativity scale to Windows . . . linux is always changing, and “stuff” does happen with it. A couple years back I had a bunch of problems with TW, but the forum guys walked me through it. And, those problems rolled over to Leap . . . forum helped again . . . and then it went on to Slow Roll, which one day just vaporized itself, nothing but an ER shell that said, “Help me” was seen . . . no help from the forum could resuscitate it.

For the Nvidia card I use “nouveau” or “default,” not the proprietary drivers, which Nvidia doesn’t keep pace with the development cycles of linux . . . .

Point of this post . . . if you have been playing with linux enough to get to TW, you should use “advanced” or “custom” installer to set up your bootloader partition, a “/” partition, and a separate “/home” partition . . . and a swap . . . because, as several others have mentioned, that way, if and when something “happens” to the / filesystem, it’s often faster to run a fresh install than to spend time parsing the problem . . . all you have to do is re-use the user name of the existing /home file in the fresh install process and all of your preferences are ready to go, all of the user files, intact . . . no need to reinstall them, etc.

For a few systems I use the same /home partition of say 100+ GB, with a couple or three user directories in it. Each / filesystem has roughly 50GB partition . . . I’m still using ext4 formatting for / and /home.

TW does have what I call “balloon” upgrades, where there will a huge amount of packages to upgrade . . . but, for the most part, with “rolling” distro, once you have it set up, no need to install an upgrade, or edit your sources . . . it just rolls on with the latest, freshest horsies . . . .

2 Likes

Wow. I forgot about all the 70’s stuff. I started my career with Assembly lang (very painful). Then to C /C++, then Java.

My first distro was SLS, then moved to Slackware in 1993 (I suspect many folks here don’t know the name with experience, some might have been in Jr High school :slight_smile: )

2 Likes

Well, everyone has their own experience and advice based on that:

  • TW: Yes, it’s the one I’ve been using for more than 5 years and what I recommend, as long as it’s a couple of PCs that you’re going to have to manage because, indeed, if you have to do a ROLLBACK, it’s very simple, but if you have to do it on more than 5 computers … it’s not so much fun.
  • LEAP: If you come from GNOME, you can stay with LEAP, because you are one of those who don’t ask much from the windowing environment. But if you need Plasma, of course, TW is the version to choose and you will always have the latest Plasma and its framework, without having to mix and match repositories in order to have the latest Plasma version.
  • Separate HOME partition: A MUST if you value your data, but yes, with EXT4 format. It is clear that none of those who recommend XFS formatting have dealt with the difficulties of recovering data from an XFS partition. Moreover, the free XFS recovery tools are so basic that you don’t recover anything.
  • Backup: GDISK (GNOME DISKS), Foxclone and Clonezilla
1 Like

Back then we had “Basic” and “4Tran” . . . we had to pick one of them to use to make our punch cards. I picked Basic, because I thought it would be easier to do stuff, but I think I learned that 4Tran might have been “better” or more simple . . . .

I wasn’t really connecting with the technology of that day . . . other than the phone receiver placed in a cradle to make the dial up internet connection . . . . All very slow and very time consuming to do the most Basic commands . . . . The math geeks were ecstatic about it . . . in their element . . . similar to “the jocks” but about another topic.

1 Like

I started using XFS back around 2002 when I was using Gentoo (talk about a painstakingly laborious distro! ). I’ve been using XFS since then (so, about 22+ years), with zero issues.

And no, not just on our personal machines at home. I worked as a developer for the 3rd largest county (supporting all departments) in the U.S. - their huge array of Unix machines all used XFS. And other companies I worked for afterwards (all world-wide software companies), also used XFS on their machines.

Maybe you’re referring to ZFS? That one is problematic for folks. I don’t think openSuse even offers ZFS (?) on any of their platforms, as well as many other distros.

EDIT: For me, I don’t state “folks should use ZFS” … I only mention we use it for our separate /home partition, so maybe other folks are pushing XFS use. :+1:

There are excellent GUI tools that use Borg as the underpinning backup method, including Pika Backup (GNOME Circle) and Vorta (fits a KDE/Qt environment better).

Personally, I use rsync over ssh to collect data within my local network (from laptops and other desktops) on to one desktop machine, and then use Vorta to create the incremental/snapshot backups from there.

2 Likes

As I said at the top of my commentary, everyone narrates their own experience. XFS is the format I used for several years on my ‘/home’ partition. ZFS, as you say, is not available on openSUSE. Until I had an unexpected crash and from that moment, the desperation was absolute in the recovery of information from XFS. Not only does it have very basic recovery tools, but the ones that actually do anything are less than 3, all paid tools, like R-Studio, which is one of the few that can barely recover XFS partitions.

If you tell me about your personal experience with array systems, it is clear to me that they will have a RAID system or at least a backup system efficient enough so that the user will not even know that a hard disk failed.

This is why I insist that for the ‘/home’ partition, the ideal is EXT4, especially because there is a much better chance of recovering the information than from a disk with XFS.

1 Like

Hello all,

ZFS is available, but I don’t know who maintains it. My understanding is that ZFS can not be an “official” part of any gnu/linux distro for licencing reasons.

https://en.opensuse.org/OpenZFS#ZFS_Best_Practices_and_Tuning

https://download.opensuse.org/repositories/filesystems/15.6/x86_64/

I have used ext4 almost exclusively since the retirement of reiserfs, RIP. Four or five years ago, I ran btrfs on top of a mdraid level 1 device for a few months, where it proceeded to become irreparably corrupted several times. I know some folk claim that btrfs on top of mdraid is ok, but my experience says otherwise.

I have a large movie and music video collection that I store on a separate filesystem on a software raid 10 far 2 device. A couple of weeks back, I switched it to XFS. For a collection of mostly huge files, XFS is definitely faster than ext4. I am ALMOST looking forward to the next power failure to see how it fairs :smile: . I also seem to loose about 250GB of usable storage on an 8TB filesystem. I have yet to find the XFS equivalent of “tune2fs -m 0.5”.

I am interested in ZFS because “checking” the integrity of this 8TB raid device takes more than twelve hours, and … ZFS seems really COOL!