It's been a while but for me personally, the number one thing that saves your ass in event of failure is installing both the up to date and the LTS kernel. If you fail to boot, it's gonna have something to do with a kernel update 90% of the time, in which case you just boot with the LTS kernel. Just about anything else can be fixed either while you're up and running, or with a live distro installed on a USB.
i think that sort of issues will no longer be relevant, considering that there are official semi-opensource nvidia-open drivers and that even wayland supports nvidia gpus just fine nowadays.
I know people who've driven drunk for more than a decade, and only crashed when they were sober. Does that mean drinking and driving is safe?
I literally had one of my devices fail to boot after an update pretty much every month, and another a few times a year. Sure, I was able to fix it with only arch-chroot, but when I switched to btrfs I didn't need to carry around a flash drive with me when updating the system...
The last few years I'm on nixos, and while it sometimes fails to update, it has not once failed to boot.
There's nothing specially destructive about dd. It's just a program that opens a file, and if you tell it to open a device file and it's got permission to write to it, it'll write to it.
You can just as easily overwrite a disk's contents with anything else that will open a file and write to it.
# cp /dev/urandom /dev/sda
will wipe your drive with random data in the same way that
mkfs.ntfs is the silent killer, other mkfs don’t allow writing over the root partition but summoning this kind of Windows magic will happily fuck up your shit.
I quadruple check and then I still don't run it because I'm too nervous so I go and do something else instead and the item remains buried on my to-do list indefinitely.
Lol, as someone who tried a bunch of different distros, I used the dd command a lot without ever realizing what could go wrong with it. I must have used up all my luck with that though, because somehow I ended up on manjaro of all things.
My last Arch death was updating my 128GB laptop after a year and breaking it with partial updates. (I’m better at partial updates now)
It also made me want to, move away from having separate /home partition. It was somewhat sad, since it was the Arch that I installed on the previous laptop and just moved the SSD when I got a new one
Where are those people whose arch installs break after every second update? I don't think I had a single problem with arch that I wasn't responsible for. Hell, I had more problems with mint on my girlfriends PC than the arch machines that I'm running.
If you don't count my 6 years with Manjarno, I've used Arch for ~7 years and only 3 times I had to fix something, 2 times because of Nvidia, once my fault. I had way more problems with Ubuntu in the past than Arch by the way. :)
I actually did this at one point, because OneDrive was the cheapest cloud service I had easy access to, back then. I did use only an encrypted vault, though.
OneDrive used to be pretty good on Windows before it started taking the documents folder hostage. Maybe needless to say I've never attempted to use it on any of my Linux installs
I tried once, could I get it work on Mint (I could not). Not that I wanted to use it at that point, I just got curious is it actually possible. Apparently yes, even though I did not succeed.
I honestly can't think of a thing I'd want to backup. Game saves, resumes, media. Its all replaceable and not worth backing up. I dont think I have a single "important" document or picture or anything.
I remember daily driving Gentoo unstable in the early 2000's on my rig in college. It was like trying to beat a Dark Souls boss every time I wanted to do an update.
Use a modern filesystem like btrfs or even lvm with thin provisioning to solve that problem. Separate partitions for ease of backup, but they all pull from the same pool of storage.
On my PC and all my home servers I keep my root partition at 10GB and haven't had any issues (although I keep everything that isn't hardware related or a system utility in home).
On my PC I crammed arch, all of KDE (+extra KDE utilities), and some other system stuff like LACT and distrobox in the 10GB root partition and run almost everything else as a flatpak installed to home.
On my servers I do the same thing but all my services are in podman containers in /opt.
Having all the software needed to make the machine work be isolated from my custom home junk makes me feel better, and I like that I can mount root as read-only to make it immutable
I think that there's a reasonable argument for a separate home partition on a multiuser system. Means that users can't fill up root.
But a lot of systems aren't multiuser these days.
EDIT: Honestly, my biggest partitioning irritation these days is that the Debian installer makes it a pain in the rear to have a larger /boot if you're also doing an LVM-based install; the default is unreasonably small these days.
Yeah i never bothered with separate /home because it feels like such a balancing act to decide on size compared to root. Especially when you use nix/guix, it would be annoying to have a small root with the store filling up over time and having to do garbage collecting often. Also, i just back up whatever files i want to make sure i don't lose, and when i do a clean install i actually want everything to be wiped anyway, including home.
zloubida explained it with the link well enough. More importantly for me, the package repository seemed to stop being maintained, because many of my packages had no updates for a significant amount of time (and they had been updated elsewhere when I checked).
Conflicts between the project leadership and the developers. I stayed on Manjaro though, this kind of things is in fact good news: it means that people in charge care, and I think things will be better thanks to that. I'll just follow the devs if things don't go better.
Wherever people say "Manjaro just broke" it's good odds they tried to use it like an Arch install. Installing drivers "the Arch way", pointing the repos at Arch, using bleeding edge kernel versions etc.
It may be Arch but it's been set up in a certain way and if you mess with it it won't work. Not messing with it is the whole point of the distro.
It does have a wiki and forums etc. so I feel that if people read a bit before doing stuff we wouldn't be having this conversation.
what's a snapshot, new kernel? lets roll!! almost 2k packages in
never really broke my arch tbh, its been serving flawlessly for 7 years now. its a decently recent thinkpad tho so stuff tends to just work and i have no gpu to bother about
Arch has never "broken" on me through my 3 years of using it, but after a while weird issues would start to appear. Like for example my login manager just stopped working permanently and I had to do startplasma-wayland. Ssh agents also subtly broke. Stuff like that. I forgot to enable fstrim and my system freaked out during I/O eventually. These are all technically user error, but I don't feel like spending so much time on managing my operating system.
I ended up switching to Fedora and was pleasantly surprised how well it's setup by default.
idk Arch's timeline, but back in Janurary CachyOS changed from using SDDM to Plasma Login Manager and their news release had some manual commands to enter in to migrate more smoothly.
I think some distros have a systemd service that does it weekly or monthly which is usually enough but in rare cases it's not, just know that if you suddenly get out of space errors on BTRFS but you're not actually out of space that's probably the issue
you can use it to (sometimes) transplant your personal config and documents onto another OS, but it comes with the tradeoff of losing space in root that might be used for massive libraries (compute libraries suck in this regard).
And sometimes your /home will depend on things that your rootfs doesn't have, which has caused problems for me before. But if you planned for these things beforehand, and your installed software doesn't do something really dumb, it can work out pretty well, although my personal experience with this is limited to effectively embedded Linux where this is usually a non-issue.
Nope. The stuff I need on my machine is all managed with git. I'd lose a few documents, but nothing I couldnt love without. Photos and important stuff are in a few different places, but not my main PC.
the one thing i wish i knew when i first installed linux was to put /home on another drive (i do symlink downloads and anything else now but i wish i did it on my desktop when i first switched)
well my desktop has the main ssd where i installed it too but that is pretty small while the second drive is something like 2tb so that's where i would have put it. its honestly not bad now i had a brief panic where it ran out but ive symlinked a lot in /home over to the new place its just wish i understood the advanced partitioning when i first set it up that's all.
You say it's not too difficult, but, I guarantee you, if I tried to do this, I'd lose not only my /home, but also my game drive. I'm that kind of dumb.
Btrfs and snapper. Only time I ever had to reload a snapshot was because of my own actions. I've been fortunate to never have any update related breaks. Maybe a couple services had hiccups, but never breaks.
The ideal system probably is something like btrfs snapshot, then restic or similar of that snapshot to a remote disk.
Btrfs doesn't dedup identical data at different locations. Restic (or anything similar non-filesystem-level) cannot take an atomic snapshot. If you have something like a PostgreSQL database, where the DBMS ensures that what's on disk is valid at any point in time, but only to an atomic view of the thing, you probably want an atomic snapshot of it.
I dunno if anyone has a single off-the-shelf utility that does both steps together.
Zfs does atomic snapshots and also computes the diff between snapshots when you zfs send/receive to remote. You can set up syncoid scripts to do all of this automagically and ensure source and dest match, have regular snapshots, and to manage retention schhedules on both ends.
you know after installing god knows how many distros and reinstalling just the same it's like easy peasy to get anything installed now and up and running/configured the way I like it within an hour. Especially if you git repo your configs. I have my Suckless suite of stuff in a repo so when I install a new distro or reinstall something I just clone that repo and i'm good to go. all my important data is backed up with borg. it all works across any distro and even BSDs.
Git repo your configs/DEs/WMs folks. such a massive time saver.
Eh, I've had worse. I'd actually like it if it included FDE (my spare Artix laptop has this config, while my main is stuck with home encryption (on seperate partition)and Suse TW, which is kinda chaotic. I complained about openrc being a pain to daily, but Artix actually makes it good and feels actually stable.
It's not so crazy these days. In several years of running arch, only once has an update broken anything (and that's because I was messing around with virtual machines).
ROFL Like, this was me on other distros just using ext4, no timeshift (made manual backups though, not as much as I should've). Now, I am understanding the hype with btrfs, Snapshots, as they have saved my ass so much. Using Garuda Linux, BTW! Sometimes a machine doesn't matter so much so...I lived life dangerously, and paid for a particularly low cost for it.
Ridiculously easy to use and quickly get back into a working system! I used it first with openSUSE Tumbleweed, after the adapting to how it functions, it was easy to use. Something I cannot live without. Edit: I had to change some text because those 4am thoughts weren't lining up right.
Finally? It's been SUSE's default for at least a decade now. It's the default for Fedora since 33, so ~5 years ago. Btrfs has been working for a long time.
Yeah and it's awesome. I have it making backup snapshots automatically and I copied the steamos-readonly script from valve to automatically lock everything except my home/root/var/opt etc (I realize you can remount ext4 as ro too, but still btrfs handles this well).
I recently swapped the contents of two drives of different sizes (one was a server and the other was a PC) and btrfs made the filesystem resize and expansion parts easy.
I was introduced to it by a synology nas a decade ago, and it's my preferred fs. I wish proxmox supported it, my compute server had to go with ext4 since it's got one disk for the system, and one for the VMs, no redundancy. They say it's in 'tech preview' but most mainstream distros have had it for years now...
Timeshift has indeed saved my distro's ass a few times. But is not a silver bullet. In the end I keep most of my stuff on an external HD and in there a folder with some notes and data about my distro configuration. If he dies (and he did) I'll bring him back.
That was me when I first started using Linux 10 years ago. I didn’t know any of this stuff, and most of it didn’t even fucking exit. Wing it and worry later. Moral of the story, I reinstalled several times a month. Lmao
194 replies
Fuck it, we ball. 3 years no arch deaths. Windows is a fuck. 1,000,000 pacman warnings.
I don't remove orphan packages, I don't setuo snapshots, I don't read arch news.
I
sudo pacman -Syu. And if I can't boot I don't do itIt's been a while but for me personally, the number one thing that saves your ass in event of failure is installing both the up to date and the LTS kernel. If you fail to boot, it's gonna have something to do with a kernel update 90% of the time, in which case you just boot with the LTS kernel. Just about anything else can be fixed either while you're up and running, or with a live distro installed on a USB.
How do you mismanage your system so bad that you need these things?
Hot take: I used arch for over a decade. It never broke unless I broke it.
By all means, back up your files but rolling release distros aren't made of arsenic, you'll be fine.
I think it's very much dependent on your hardware.
I had some issues with my Nvidia drivers getting messed up when the card was relatively new. I haven't had any issues in the past two years though.
i think that sort of issues will no longer be relevant, considering that there are official semi-opensource nvidia-open drivers and that even wayland supports nvidia gpus just fine nowadays.
I know people who've driven drunk for more than a decade, and only crashed when they were sober. Does that mean drinking and driving is safe?
I literally had one of my devices fail to boot after an update pretty much every month, and another a few times a year. Sure, I was able to fix it with only arch-chroot, but when I switched to btrfs I didn't need to carry around a flash drive with me when updating the system...
The last few years I'm on nixos, and while it sometimes fails to update, it has not once failed to boot.
Thats's a really bad analogy
One endangers others while the other makes a couple of files and photos go poof
The danger and the stakes are not what is being compared, though. It's a perfectly good analogy.
I'm convinced that majority of people using timeshift couldn't recover from their system failing to boot or a drive failure.
And if you could tell me how because I got no idea how to either.
You can just boot a Linux Mint live session and restore from there.
why mint in particular lol
Caus mint is the superior distro. I use mint btw
It has Timeshift preinstalled, because Timeshift is maintained by the Linux Mint team. You can use any live session and install Timeshift, though.
I’m here for a good time, not a long time
Also, my last Arch death was due to accidentally ovverwritimg the whole drive, so no amount of split partitions is going to save me
They call it disk destroyer for a reason. You gotta alias that shit.
alias dd='echo Did you triple check which device you picked? Of course not, dipshit. Check again'I checked five times, actually. And then I check some more.
There's nothing specially destructive about
dd. It's just a program that opens a file, and if you tell it to open a device file and it's got permission to write to it, it'll write to it.You can just as easily overwrite a disk's contents with anything else that will open a file and write to it.
will wipe your drive with random data in the same way that
will.
You're technically right, but the way most people use dd ensures its like handling a loaded gun. How often do you cp /dev/sdwhy?
mkfs.ntfs is the silent killer, other mkfs don’t allow writing over the root partition but summoning this kind of Windows magic will happily fuck up your shit.
Hahahahah I need to do that lmao!!!
Great advice. I rarely use dd, but when I do it always sends chills down my spine.
Surely double checked is enough
It's not. Neither is triple checking. Notice how the alias doesn't actually run
dd? That's on purpose.I quadruple check and then I still don't run it because I'm too nervous so I go and do something else instead and the item remains buried on my to-do list indefinitely.
Lol, as someone who tried a bunch of different distros, I used the dd command a lot without ever realizing what could go wrong with it. I must have used up all my luck with that though, because somehow I ended up on manjaro of all things.
My last Arch death was updating my 128GB laptop after a year and breaking it with partial updates. (I’m better at partial updates now) It also made me want to, move away from having separate /home partition. It was somewhat sad, since it was the Arch that I installed on the previous laptop and just moved the SSD when I got a new one
Where are those people whose arch installs break after every second update? I don't think I had a single problem with arch that I wasn't responsible for. Hell, I had more problems with mint on my girlfriends PC than the arch machines that I'm running.
If you don't count my 6 years with Manjarno, I've used Arch for ~7 years and only 3 times I had to fix something, 2 times because of Nvidia, once my fault. I had way more problems with Ubuntu in the past than Arch by the way. :)
There's the argument to be made that all three were your fault, since you decided to use nvidia on Linux.
Hyuk hyuk hyuk.
Well, I also used ATI in the past but it was way worse. :)
This is still true but I find you are using bcachefs where every update is a death threat.
Why would you want your entire home folder in syncthing?
I'm very attached to my
~/.cache!stresstest your io i guess
A live copy of my home folder and the few remaining secrets/configs i don't care to put into home manager.
I try not to put dangerous data into HM or my Nix configs, I just let syncthing whisk it away to my homelab, it has versioning turned on.
also .stignore is your friend
also, machines wipe themselves at each boot! service state data is distributed via zfs send! zfs that actually works!
Syncthing will also propagate removals or bad edits
also support versioning if you really wanted to keep them.
i mean, all the important stuff is saved to OneDrive, right?
Straight to jail.
You missed the space between one and drive… and you accidentally capitalized a couple letters.
Don’t correct me on this one. Please.
All my important stuff is saved to SomeDrive
Yes, but actually no.
I actually did this at one point, because OneDrive was the cheapest cloud service I had easy access to, back then. I did use only an encrypted vault, though.
OneDrive used to be pretty good on Windows before it started taking the documents folder hostage. Maybe needless to say I've never attempted to use it on any of my Linux installs
I tried once, could I get it work on Mint (I could not). Not that I wanted to use it at that point, I just got curious is it actually possible. Apparently yes, even though I did not succeed.
This is unfortunately me with google drive for a few things.
Pfft.. it's been fine for the last 10 years.
Seriously. I've spent more time maintaining my debian home server than I have my two arch installs combined.
These memes have always been propagated by people that decide that fucking with their OS is a good idea.
It is.
This is my setup that I have been using for years. I have never had a serious Linux update issue that could not be fixed.
Well la dee da look at this guy who knows how to fix linux!
Yeehaw!
I, too, am familiar with the art of grub rescue env!
I have seen its darkest depths.
And oh how deep they doth go!
(I use arch btw)
I live like this, it has been 10 years, no problem
Same here, not sure what they on about.
I third this
I'm stupid and I didn't know there was other ways of living
there are two types of computer users
Not me!
I honestly can't think of a thing I'd want to backup. Game saves, resumes, media. Its all replaceable and not worth backing up. I dont think I have a single "important" document or picture or anything.
It's time saved rebuilding all of it
Or those who can fix it.
what are the cheap ways of setting up automatic backups?
Btrbk if you use the right filesystem.
Rsync and cron
i did rsync and cron for years. now I switched to rsync and systemd timers 🙂🙂
Systemdeez nuts
yes, but to what medium
Whatever’s around. Some writable filesystem.
Rule of thumb I adopted: If you never update arch, it will never break.
Arch noobs after destroying their system every month be like:
Wait. Are you saying you can choose not to rawdog your os ?
I remember daily driving Gentoo unstable in the early 2000's on my rig in college. It was like trying to beat a Dark Souls boss every time I wanted to do an update.
In one of my university java courses, another student was frequently compiling Gentoo packages on his laptop during class.
Once I researched more about it I also installed Gentoo on my laptop and started building updates during class 😁
Mint users with no zfs, no per block checksumming, no syncoid, no ability to boot into prior version or revert right in the boot loader...
Even ext4 still has safety measures (BLOAT!11!!1) because of journaling.
This is why i compile my kernels on ext2 with without swap for maximum utilization /j
How do you know about me.
Non-issue. I just scatter fragments of my data across so many devices that some version of something is bound to be somewhere. ¯\_(ツ)_/¯
Ah yes, the shotgun approach to data integrity
Separate /home is such a scam. Never have I run out of home space, but boy is 27GB not enough for a root partition that runs any kind of service.
I only do it on my ancient laptop because for some reason, it makes things a little faster. I don't know why.
Use a modern filesystem like btrfs or even lvm with thin provisioning to solve that problem. Separate partitions for ease of backup, but they all pull from the same pool of storage.
On my PC and all my home servers I keep my root partition at 10GB and haven't had any issues (although I keep everything that isn't hardware related or a system utility in home).
On my PC I crammed arch, all of KDE (+extra KDE utilities), and some other system stuff like LACT and distrobox in the 10GB root partition and run almost everything else as a flatpak installed to home.
On my servers I do the same thing but all my services are in podman containers in /opt.
Having all the software needed to make the machine work be isolated from my custom home junk makes me feel better, and I like that I can mount root as read-only to make it immutable
/ size becomes an issue when you install games via pacman
symlink the directories to somewhere on another partition
I have / on SSD (less writes) and /var and /home on HDD
I think that there's a reasonable argument for a separate home partition on a multiuser system. Means that users can't fill up root.
But a lot of systems aren't multiuser these days.
EDIT: Honestly, my biggest partitioning irritation these days is that the Debian installer makes it a pain in the rear to have a larger /boot if you're also doing an LVM-based install; the default is unreasonably small these days.
Probably my installed games alone are larger than that
Yeah i never bothered with separate /home because it feels like such a balancing act to decide on size compared to root. Especially when you use nix/guix, it would be annoying to have a small root with the store filling up over time and having to do garbage collecting often. Also, i just back up whatever files i want to make sure i don't lose, and when i do a clean install i actually want everything to be wiped anyway, including home.
It's the opposite for me.
This is me with endeavourOS lmao. Anything important is on my mega/Codeberg. If it dies it dies
if it breaks, just fix it
Dotfiles in git my dude.
Edit: or even better chezmoi with dotfiles and ansible bootstrapping
my manjaro has been running the same install for 6 years on ext4 no separate home partition...
what are yall doing to brick your machines?
Right? The OP describes me (except I like to think I'm not that sweaty) and I don't remember ever having reinstalled because things were broken.
Fuckin around :)
Finding out :(
You're still using manjaro after all their drama? I jumped ship.
What happened?
zloubida explained it with the link well enough. More importantly for me, the package repository seemed to stop being maintained, because many of my packages had no updates for a significant amount of time (and they had been updated elsewhere when I checked).
Conflicts between the project leadership and the developers. I stayed on Manjaro though, this kind of things is in fact good news: it means that people in charge care, and I think things will be better thanks to that. I'll just follow the devs if things don't go better.
It still gets updates and still works so why jump ship.
I remember using manjaro vanilla and breaking it in a single month of use.
No aur enabled, no nothing - it just broke, it wasn't even funny
Wherever people say "Manjaro just broke" it's good odds they tried to use it like an Arch install. Installing drivers "the Arch way", pointing the repos at Arch, using bleeding edge kernel versions etc.
It may be Arch but it's been set up in a certain way and if you mess with it it won't work. Not messing with it is the whole point of the distro.
It does have a wiki and forums etc. so I feel that if people read a bit before doing stuff we wouldn't be having this conversation.
The funny thing is that i did none of that, the hardware i was using didn't need additional drivers and i didn't add any extra repos.
Perhaps i did something wrong, or autoupdates were interrupted but i don't remember doing anything extraordinary with the install.
Manjaro isn't Arch. Doesn't use the same repos.
to be honest that setup is faster to arch-chroot if things go wrong.
I always back up anything that I feel is actually important, do people actually rely on the stability of their OS?
Not that Arch has caused me any issues whatsoever so far, going on 2 years now while very much a Linux noob.
Just get your snapper to run before/after any update and if you bork something, roll back. Easy peasy
what's a snapshot, new kernel? lets roll!! almost 2k packages in
never really broke my arch tbh, its been serving flawlessly for 7 years now. its a decently recent thinkpad tho so stuff tends to just work and i have no gpu to bother about
I'm rawdogging my Debian Unstable install
Nah. We just know how to boot a broken system and fix it.
Exactly. Nothing of importance is stored solely there anyway.
And for me I need every single GB for installing games, no space left for backup management.
I run a live real time dist straight into ram. I communicate with Linux telepathically.
Arch has never "broken" on me through my 3 years of using it, but after a while weird issues would start to appear. Like for example my login manager just stopped working permanently and I had to do
startplasma-wayland. Ssh agents also subtly broke. Stuff like that. I forgot to enablefstrimand my system freaked out during I/O eventually. These are all technically user error, but I don't feel like spending so much time on managing my operating system.I ended up switching to Fedora and was pleasantly surprised how well it's setup by default.
idk Arch's timeline, but back in Janurary CachyOS changed from using SDDM to Plasma Login Manager and their news release had some manual commands to enter in to migrate more smoothly.
every login manager was broken, it'd just crash after I put in my password. It was the final straw
Honestly agree, it's always the little issues that are problems.
Me with arch: what is timeshift? Time to updstr without backup, lol!
*BTRFS users when you don't rebalance soon enough and run out of inodes so you can't boot your computer
How often do you need to rebalance?
When the moon is in the Seventh House and Jupiter aligns with Mars,
then peace will guide the planets, and love will steer the stars.
I think some distros have a systemd service that does it weekly or monthly which is usually enough but in rare cases it's not, just know that if you suddenly get out of space errors on BTRFS but you're not actually out of space that's probably the issue
my current openSUSE install with btrfs has lived for about 6 years now and I haven't done any manual maintenance on btrfs
meh, just backup home.
I check all these boxes. This install is ancient now. I moved the disk into a new PC (Intel/nvidia to amd/amd) and it didn't even stutter.
I had issues in the past, sure, but nothing really severe. Archlinux.org frontpage has information on breaking updates if need be.
I will simply reinstall and forsake my files if it shits the bed catastrophically
Timeshift? BTRFS Snapper layout!
My /home/$USER partition is on a separate disk. Me wicked smaht
SMRT!
I ran linux with ext2, no timeshift, and home on the root partition. Back then when I ran the machine on kernel 0.97.
A separate partition for /home/
Why haven’t I thought of this before??
you can use it to (sometimes) transplant your personal config and documents onto another OS, but it comes with the tradeoff of losing space in root that might be used for massive libraries (compute libraries suck in this regard).
And sometimes your /home will depend on things that your rootfs doesn't have, which has caused problems for me before. But if you planned for these things beforehand, and your installed software doesn't do something really dumb, it can work out pretty well, although my personal experience with this is limited to effectively embedded Linux where this is usually a non-issue.
Yeah but you do backups right? Right?
I dont actually but I can reinstall the system easily, and the user data is mapped to my nas drive.
Sure if I would delete all my files on the nas by mistake, I guess I lose them. But hasn't happened.
Should actually back up my nas... Will get some external drive to do that I guess.
Nope. The stuff I need on my machine is all managed with git. I'd lose a few documents, but nothing I couldnt love without. Photos and important stuff are in a few different places, but not my main PC.
I'm in this picture... My build had been stable for a year and I only once had my PC booting in safe mode
I'm in this picture and I don't like it.
More like "I'm in this picture and I like it."
the one thing i wish i knew when i first installed linux was to put /home on another drive (i do symlink downloads and anything else now but i wish i did it on my desktop when i first switched)
Are you meaning on a different partition or a completely different hard drive?
well my desktop has the main ssd where i installed it too but that is pretty small while the second drive is something like 2tb so that's where i would have put it. its honestly not bad now i had a brief panic where it ran out but ive symlinked a lot in /home over to the new place its just wish i understood the advanced partitioning when i first set it up that's all.
Not too difficult to fix this in post.
You say it's not too difficult, but, I guarantee you, if I tried to do this, I'd lose not only my /home, but also my game drive. I'm that kind of dumb.
dies
"Welp, time to distrohop."
Btrfs and snapper. Only time I ever had to reload a snapshot was because of my own actions. I've been fortunate to never have any update related breaks. Maybe a couple services had hiccups, but never breaks.
Old man meme. It's like saying "you need the command line for Linux!"
Decade running arch on a laptop w encrypted drives, the only fails were from powerdevil.
Important files go to repos / backup disks. Workstations are sinners in the hands of an angry god.
The ideal system probably is something like btrfs snapshot, then restic or similar of that snapshot to a remote disk.
Btrfs doesn't dedup identical data at different locations. Restic (or anything similar non-filesystem-level) cannot take an atomic snapshot. If you have something like a PostgreSQL database, where the DBMS ensures that what's on disk is valid at any point in time, but only to an atomic view of the thing, you probably want an atomic snapshot of it.
I dunno if anyone has a single off-the-shelf utility that does both steps together.
Zfs does atomic snapshots and also computes the diff between snapshots when you zfs send/receive to remote. You can set up syncoid scripts to do all of this automagically and ensure source and dest match, have regular snapshots, and to manage retention schhedules on both ends.
No sweat. I prefer to reinstall and start fresh every now and then anyway.
'It has been
20 days since the last reinstall'you know after installing god knows how many distros and reinstalling just the same it's like easy peasy to get anything installed now and up and running/configured the way I like it within an hour. Especially if you git repo your configs. I have my Suckless suite of stuff in a repo so when I install a new distro or reinstall something I just clone that repo and i'm good to go. all my important data is backed up with borg. it all works across any distro and even BSDs.
Git repo your configs/DEs/WMs folks. such a massive time saver.
I feel attacked! (but I have /home separate; and backups of everything except big game installs I can easily reinstall)
Eh, I've had worse. I'd actually like it if it included FDE (my spare Artix laptop has this config, while my main is stuck with home encryption (on seperate partition)and Suse TW, which is kinda chaotic. I complained about openrc being a pain to daily, but Artix actually makes it good and feels actually stable.
You can backup via a live enviroment anyway.
It's not so crazy these days. In several years of running arch, only once has an update broken anything (and that's because I was messing around with virtual machines).
ROFL Like, this was me on other distros just using ext4, no timeshift (made manual backups though, not as much as I should've). Now, I am understanding the hype with btrfs, Snapshots, as they have saved my ass so much. Using Garuda Linux, BTW! Sometimes a machine doesn't matter so much so...I lived life dangerously, and paid for a particularly low cost for it.
Snapper is sooooo nice 🤩.
Ridiculously easy to use and quickly get back into a working system! I used it first with openSUSE Tumbleweed, after the adapting to how it functions, it was easy to use. Something I cannot live without. Edit: I had to change some text because those 4am thoughts weren't lining up right.
xfsgang? 🥲This is me!
I wouldn’t call every 2 years "constantly"
You update arch every 2 years?
I made a /home partition on eOS at least! But given that the drive is only 128G, I don't store much in it that isn't synced from desktop.
The problem with btrfs is no installer I've seen supports reusing an existing subvol for your home partition. Not that I can figure out at least.
Could you set a 1GB home partition on install and change it in fstab once install is complete and you're booted up?
💀
Been there, done that
btrfs with pacman hook and i do backups semi reguralry (monthly + -)
This is me... IDGAF
I don't keep anything important on my primary daily driver. I keep it all on my NAS and that has backups.
I was one of them few years ago...
😎
At least they're not waiting several months for accumulated updates! I've been told that's how you break an Arch system.
Not really. Might have to update the keychain first or update the package mirrors of it's a very long time but it's not a huge deal
pacmanupdates the keyring first now. It’s gotten quite difficult to break Arch.does btrfs finally work now?
Finally? It's been SUSE's default for at least a decade now. It's the default for Fedora since 33, so ~5 years ago. Btrfs has been working for a long time.
Yeah and it's awesome. I have it making backup snapshots automatically and I copied the steamos-readonly script from valve to automatically lock everything except my home/root/var/opt etc (I realize you can remount ext4 as ro too, but still btrfs handles this well).
I recently swapped the contents of two drives of different sizes (one was a server and the other was a PC) and btrfs made the filesystem resize and expansion parts easy.
I was introduced to it by a synology nas a decade ago, and it's my preferred fs. I wish proxmox supported it, my compute server had to go with ext4 since it's got one disk for the system, and one for the VMs, no redundancy. They say it's in 'tech preview' but most mainstream distros have had it for years now...
I think I had the option for btrfs on Proxmox (but I did choose ZFS for simple replication to my NAS)
I have ext4 and dont use a separate3 home partition but I use timeshift
That's me right now. Just use deja dup on my home to my NAS.
Meh, I keep everything important on a separate drive altogether.
Timeshift has indeed saved my distro's ass a few times. But is not a silver bullet. In the end I keep most of my stuff on an external HD and in there a folder with some notes and data about my distro configuration. If he dies (and he did) I'll bring him back.
That was me when I first started using Linux 10 years ago. I didn’t know any of this stuff, and most of it didn’t even fucking exit. Wing it and worry later. Moral of the story, I reinstalled several times a month. Lmao