You are not logged in.
Hello everyone,
I’m relatively new to Arch Linux and have been learning a lot from the community and the documentation. Thank you for all the work that goes into maintaining such an incredible distribution.
I wanted to ask about the habits or “rituals” experienced Arch users follow before performing a full system update.
For example:
Do you check specific documentation or news pages before updating?
Are there commands or checks you typically run first?
How often do you update your system?
Do you maintain any kind of backup or snapshot workflow before running pacman -Syu?
I’m trying to develop a safe and consistent update routine, and I would really appreciate learning how more experienced users approach this process.
Thank you very much for your time and guidance.
Best regards,
Santo
Offline
I update about once a week by blindly running pacman -Syu. If something goes off the rails, I then read the front page of archlinux.org for announcements. As a moderator who spends a significant amount of time on the forums, I often see foreshadowing information before I get bitten. This happens maybe once or twice a year.
After that, I use auracle outdated to check for AUR packages that need updating.
The only thing I ever backup are my own work products. And they are thoroughly backed up. As to operating system files, no backups. I can start fresh and rebuild a working system from scratch in less than an hour, so why?
Occasionally (maybe monthly) I run pacman -Qdt to see if there are any orphaned implicit dependencies that are no longer needed. If there are, I remove them.
Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
The shortest way to ruin a country is to give power to demagogues.— Dionysius of Halicarnassus
---
How to Ask Questions the Smart Way
Offline
I update about once a week by blindly running pacman -Syu
this is true, but the only difference is that checkupdates is used from pacman-contrib, and doesnt update once a week.. more like whenever mirrors start to 404
pacman will throw an error anyway if theres something it cant handle, then you can go check the news
The only thing I ever backup are my own work products. And they are thoroughly backed up. As to operating system files, no backups. I can start fresh and rebuild a working system from scratch in less than an hour, so why?
backups are overrated indeed, on the server realm though before transferring everything you would want to backup a lot of stuff (although just clone the server at that point)
arch rarely breaks anyways, the only breakages that really happen are manual interventions
After that, I use auracle outdated to check for AUR packages that need updating.
this one has a combined script that checks both the aur and main repos by using aur-vercmp and checkupdates from pacman-contrib
#!/bin/sh -eu
/usr/bin/checkupdates &
pacman -Qm | aur vercmp &
waitLast edited by system72 (2026-03-08 07:12:43)
Offline
I ignore kernel updates and update them explicitly when I intend (or have) to reboot anyway and I'd suggest to have two kernels and tic-toc update them (ie. the second one only after assessing that the first one didn't cause a problem) to always have a working kernel (so you don't have to rely on live systems for offline fixes) - and then I generally check whether important™ applications still work after the update before im- or explicitly cleaning the pacman cache (to not having to rely on the ALA)
But I frankly can't recall when I was facing any sort of problem.
Subscribing to the news is certainly a good idea and keep in mind that every gnome update will typically break all your 3rd party extensions (respos, AUR or gnome-look.org or whatever, doesn't matter), so you would want to check their update status before updating gnome.
Online
I just "pacman -Syu" whenever I feel like it. I'm cocky enough that I believe I can fix any problem that I run into.
Offline
I update daily when I shut down my PC. I use informant to get notified about news items beforehand and then decide if I need to deal with the change right then, else I postpone to next day.
I have a live-iso handy in case something breaks, and reflash that every few months.
I also tend to not overbloat my machine and don't use a DE, so the likelihood of something breaking horribly are pretty slim. No fancy setups either, just plain ext4 on lvm, no dmraid or btrfs. I have maybe one borked boot a year, if that.
Offline
I run a script that checks for news, updates the system and optionally AUR updates, it also has the option to keep the terminal window open if something odd happened so I can check it. I have an applet checking for updates every 30 minutes, which is probably too often ;-)
Rlu: 222126
Offline
Do you check specific documentation or news pages before updating?
I read what aurman forwarded from RSS feed. (Yes, I still use aurman).
How often do you update your system?
In average between daily and weelky.
Do you maintain any kind of backup or snapshot workflow before running pacman -Syu?
I read output after update for anything unusual, for example .pacnew / .pacold / .pacsave files installed. I merge them using this helper launcher:
#!/bin/bash -xe
pacsrc="$1"
pacdst=$(sed 's/\.pac\(new\|old\|save\)$//' <<< "$pacsrc")
if [ "$pacsrc" == "$pacdst" ]; then
echo "Files are the same"
exit 0
fi
meld "$pacsrc" "$pacdst" 2>/dev/null
rm -i "$pacsrc"Also usually after one host updated I update another one because have configured cross-cache.
Offline
I have checkupdates bound to a user service that runs every few hours.
If there are any updates dunst shows them, then I decide if I want to update or wait.
AUR updates are done by a script and manually, every now and then or if a rebuild is necessary!
When useful I backup parts of my HOME directory, but, no system parts.
I always forget to read the news;) but, breakage has been years ago, can't even remember!
Offline
pacman -Syu and aur sync -Su (aurutils) every morning. If some AUR package needs rebuilding (rebuild-detector) then I rebuild it.
If something breaks then I reboot.
If it is still broken then I read the news.
If it is still broken then I try to unbreak it.
If it is still broken then I go to the forums.
How it feels to run shred/wipe in a COW system
Offline
I use the Apdatifier widget in KDE Plasma to regularly check for updates. I usually update about twice a week and reboot only if necessary. Apdatifier has a news reader and I also use informant to show latest news. I usually just click the "upgrade" button in Apdatifier to upgrade my system, which runs yay -Syu and also handles flatpak and fwupdmgr updates too.
This setup has not failed me once in the ~8 months I've been using it and it makes managing my systems so much easier since I have a constant reminder to upgrade my systems.
I only do a full reboot if the kernel is updated or an initrd rebuild is triggered. Otherwise I use systemctl soft-reboot if any services need to be restarted as it's much quicker
Last edited by EJSnow (2026-03-08 17:26:22)
I hate the rabbit hole I have fallen down but the sunk-cost fallacy, pacman, and the AUR prevent me from picking up and moving to Fedora or something.
Offline
Well I'm not a experienced user yet, but here is my workflow, first thing I have AMD iGPU and no NVIDIA/dGPU, so each updates kinda feels relaxing, like usually ntg breaks (except the TLB Flush error terrors me)
best practice is, read the news, before major kernel or package updates, lurk and skim the internet "if" any widespread breakages
keep snapshot/timeshift at least every 2 months
and normally update pacman/AUR, flatpak, nix etc, in every ~48 hours, It is a best practice I guess.
and reboot for voodoo reasons
and if everything works? Chill
Edit:
yes, I also recommend having another working system/OS or Live OS just in case.
Last edited by 5hridhyan (2026-03-08 17:54:16)
---
Offline
I use a script to wrap my updates which first checks if there is any new Arch news since my last update. It can be run from the command line or via a mouse click via the included desktop launcher.
Offline
Hello everyone,
Thank you all for replying to the thread so quickly. It’s really great to see this kind of support from the community.
It was also interesting to read the different approaches people take. Some experienced users seem very relaxed about updates, while others follow a more careful and structured routine. Seeing both perspectives was very helpful.
I think I will try to adopt a balance between the two approaches — staying reasonably cautious while also keeping things simple.
Also, I’m curious to hear your opinions about using Snapper with Btrfs. I configured my system with Snapper snapshots as an additional safety layer before updates, but I noticed some people prefer simply rebuilding the system if something goes wrong. I would be interested to know which approach you personally prefer and why.
Thanks again to everyone who shared their workflow and advice. I really appreciate it.
Offline
I update every day. I check arch homepage before update. I wrote my own update script which uses systemd system-update.target to install updates in a separate reboot in a kind of minimal environment. I update kernels using separate booted kernel (so I update linux-lts when booted in linux and vice versa) as a separate step. Didn't experience any issues with that approach so far. My FS is ext4, I have some irrational mistrust towards btrfs. Was going to try bcachefs as it matures, but it seems that for a foreseeable future I'll stick with ext4.
Offline
I don't do anything special except for:
1) weekly root fs snapshots (btrfs) and backup of efi + boot partitions, I keep 4 of them in rotation just in case, and backup those to NAS and external rescue drive;
2) versioning /etc (using git, but other options like hg are viable) before every update (if there were changes) and after any manual configuration changes and/or cleanup;
3) weekly (unless there is an immediate issue) /etc cleanup - like diff configs with pacnew files.
Usually, I update daily, with the exception of packages like python (.XX updates), which require rebuilding a large number of AUR packages (mostly automatically using a script, but unexpected situations can arise, so it may take some time).
There is a tool + pacman hook that will force you to read arch news before updating:
https://aur.archlinux.org/packages/informant
It’s a good practice to keep an eye on certain boot-critical packages (depending on your configuration) - in that case, taking an extra snapshot (or doing a partial upgrade until you have the time to fight possible problems; but be sure you know what you are doing) can be a good idea. Of course, it’s possible to do snapshots before every update, but I think it’s overkill.
It’s pretty safe that way. The only issue I had in the recent couple of years is the mdadm upstream bug (4.4→4.5-1, fixed in 4.5-2). But as the mdadm in my “critical package list”, I rebuilt initramfs only for one kernel, and it was easily fixed by booting another kernel.
Offline
I have btrfs snapshots setup via snapper and I use snap-pac to automatically take a snapshot of my root on every pacman update. Because I have this safety net, I mostly just pacman -Syu every single day without a care in the world.
I've rarely had to use the snapshots though. Most of the time any breakages can be fixed with a simple package downgrade using the downgrade tool from the AUR.
I am subscribed to the Arch Linux subreddit so I usually get a heads up if anything pops up on Arch News, but you can definitely subscribe to the RSS feed.
On my NAS, my only precaution is just to use the LTS kernel, but that's more to make sure ZFS doesn't break during kernel upgrades than anything else.
That's it really. I think people make too big a deal out of Arch updates.
Offline