They see us Cinnamon Rolling, they're rating: GeckoLinux incorporates kernel 5.16 with familiar installation experience Most distros haven't got to 5.15 yet, but openSUSE's downstream project GeckoLinux boasts 5.16 of the Linux kernel and the latest Cinnamon desktop environment. Some of the big-name distros have lots of downstream projects. Debian has been around for decades so has umpteen, including Ubuntu, which has dozens of its own, …
"...which makes it easy to roll back breaking changes."
Specific files or the entire disk? 10 days later... roll back to 10 days before? If it's all or nothing, daily snapshots without the requirement of any specific FS works just as well, if not better considering it might take considerable time to rebuild. Putting the daily snapshots on an external FS that has some kind of matrix/reconstruction support seems like a winner.
P.S. Reiser was pretty bad. I still have a ~2006 laptop (pre-Novell sellout) were the FS is borked so badly that it will at random not boot the kernel. Although I haven't turned it on in about a decade, the last time I did it took about 15-20 reboots to get it up so I could back it up for 1 last time (if I wasn't watching TV at the time, it might of been slammed on the floor a few times).
Can't speak to SUSE/btrfs snapshots, but I've tried it out with Ubuntu/ZFS 21.10 - there are hierarchical filesystems - everything system related is in partitions under rpool/ROOT and every user's home directory is a separate partition under rpool/USERDATA. Snapshots of rpool/ROOT are taken when you install/update packages, and you can rollback the system independently of the user data - you can even directly boot an older system from the snapshots without having to fully rollback to it. Incidentally, this is the main pain point with the snapshots - snapshots are cheap to take (virtually free), but rebuilding grub menus when a snapshot is taken is slow as hell.
I'm not sure if there are automatic snapshots taken of rpool/USERDATA, but a couple of features are probably missing before it makes it to mainstream. Ideally, USERDATA would be fairly continuously snapshotted, every minute or so, with the system automatically dropping snapshots so you could easily look at your files 1/2/3/4/5/10/15/30/60 etc minutes ago, with Nautilus UI integration to have a slider to automatically change your view of a directory.
Putting the daily snapshots on an external FS that has some kind of matrix/reconstruction support seems like a winner.
I know of many people (particularly in education in the US, weirdly) who use ZFS snapshots for all their servers. Each server in each school in their district sends snapshots to a remote ZFS master storage for backup/restore, and when they need a replacement server they just restore the previous snapshots centrally and then send the new server out. The central ZFS server is their hot backup, and they only need one set of LTO backup hardware rather than one in each school.
Snapshots get taken just before updates get applied and they only apply to affected subvolumes (e.g. /usr) without impacting user data in /home or logs in /var. btrfs does also allow rolling back of individual files too though. Snapshots work like subvolumes and can also be mounted, allowing you to grab copies of files. It's just like how NTFS VSS snapshots work, except the actual rollback method differs slightly due to differences in legacy cruft.
When System Restore is engaged on Windows, data needs to be copied from snapshots as part of the rollback attempt, as program and user data can be scattered about in multiple places and one does not wish to lose anything as a result. With Linux, because data is stored in stricter hierarchies, one can easily swap the /usr subvolume mount to use a working snapshot from before the RPM transaction occurred. This differs from LVM in that there's less space wastage and better performance since all relevant mountpoints can share space within a single partition using a single filesystem.
IMHO, the best approach is the one macOS Monterey uses with APFS. The base system is cryptographically signed as an immutable whole volume with updates appended to it using snapshots, allowing predictable/verifiable changes with the possibility of rollbacks. Likewise, Time Machine uses snapshots as a high-performance cache to speed up restores, while cloning their contents to external drives as incremental restore points, excluding read-only system data.
Well, I've got an FTP server still running it. My MX did too until fairly recently. It's true that a power cut killed it pretty effectively, but a twenty year old 2.5" IDE drive was probably on borrowed time.
I loved SUSE mostly because of the combination of YaST and KDE, by the way. These days I still love KDE, but can get along with Cinnamon. I've found a bunch of stuff that substitutes for YaST, kind of, but DEB really kicks RPM, IMO.
-A.
Yest is nice, but its no necessarily as good as at any one think. Its consistent, but most users do not need to do a lot fo what it does.Those users who do can handle multiple tools. You still end up with separate config tools for desktop environments.
OpenSuse is nice, but its higher maintance and has smaller repos than Debian based repos. You end up adding far more extra repos.
One of the things I do like about OpenSuse is Btrfs. While its true the developers recommend against using it, the Btrfs wiki page linked to says "Note that in many cases, you don't want to run fsck. Btrfs is fairly self healing" and links to this.
Im currently running Opensuse 15.3 with KVM modules for nearly 6 months ? Maybe? I dont remember when i shut it down and boot it up and keep messing around with it. It has been solid to my tampering and copying stuff and installing and uninstalling ... Compared to a similar headless with Fedora was .... Lets say erased from existence after couple of days and week of frustrating problems.
I am pleasantly surprised by the stability and availability of certain packages which were nuked from RHEL and Fedora repos.
A very long time back i was a Slackware fan but Opensuse is something i might consider a secondary goto OS for experimenting with less worry.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2023
