Or should I go 11 > 12 > 13?
Edit: Thanks for all the replies. I asked this out of laziness and apparently trying this is not a lazy thing to do. I’m not Bilbo Baggins seeking an adventure. Will go with 11 > 12 > 13 way, though might stay at 12 for a while at this point. You know, lazy. :)
As someone who went through the 12-13 upgrade last week. Do not go directly. I had to rollback twice during the 12-13 upgrade(thank you timeshift) so I couldn’t imagine trying to run an 11-13 upgrade.
It’s a headless server, so I’m thinking “how bad could it be” but I would be more cautious.
11 > 12 > 13 is way safer!
No.
By itself, apt will give you headaches.
Debian migrated to new paths for security non-free firmware in repositories from 11 to 12, and apt goes to v3 in 12 to 13, which changes the format of sources. There is a new apt modernize-sources command, but it assumes your paths are correct.
If you know what you’re doing, you can do this by correcting the repo paths and do the without-new-packages upgrade, but be prepared to fix apt.
If you’re a casual user, maybe stick with 11>12>13.
I know what I’m doing but looking at every comment here, it’s not a wise thing to do it seems. So casual or not 11 > 12 > 13 is the proper and most likely still the easiest way. It’s a good thing that I asked before doing some potentially mad thing.
Honestly, there were so many fundamental changes in the 13 upgrade for certain packages that I had to fix on a couple of machines that I’d be hesitant to try no-scoping the 11 > 13 upgrade.
I flew by the seat of my pants and managed to pull off 10 directly to 12, but I wouldn’t do it for this one.
Well, if there are issues like even in normal upgrade, it’s better not to jump on a thing like this.
Still, it’s good to know that this is technically possible, though it’s not for a lazy person who just wants to update his server. Gotta check Debian changelog.
11-12 should be well tested. 12-13 should be well tested. 11-13 may work, but you may be the tester.
I’d step through one at a time.
For fixed-release distros (Debian/Ubuntu), the upgrade path is usually sequential.
The main implication: if you skip, you’re outside the tested upgrade path. That can mean broken packages, orphaned configs, security regressions, or a system that simply won’t boot. Sometimes you can force it and it’ll work, but it’s a gamble…
Likely not. I’ve tried skipping a release once by accident (I didn’t pay enough attention) and it ended with a bricked system and a full reinstall. Don’t do it.
It seems nobody really tested 11 to 13, or maybe any kind of major-version-skipping and you won’t find direct experiences here.
Your best bet is to follow the official procedure, so 11 -> 12 -> 13. I’ll leave you with the official upgrade guide for 11 to 12 and 12 to 13.
It seems longer than it is, as not every step is actually required for every system. When upgrading VMs, a snapshot pre-upgrade can also help you skip backup-steps in the guides.
Well, certainly not gonna be a tester. If it isn’t tested, there is no need to take things further. Thanks for the guides by the way.
Major version changes for any software from the OS right down to a simple notepad app should update as sequentially as possible (11>12>13>etc). Skipping over versions is just asking for trouble, as it’s rarely tested throughly.
It might work, but why risk it.
An example: if 12 makes a big database change but you skip over that version, 13 may not recognize the databases left by 11 because 12 had the code to recognize and reformat the old database while that code was seen as unnecessary and removed from 13.
Stuff like this is also why you can’t always revert to an older version while keeping the data/databases from the newer software.
Yeah, that makes sense. Major changes in software are no joke. If it wasn’t my actual server I could’ve tried it though, could be a fun thing to pass time.
