Or asked the other way around: How long do you keep your servers running without installing any software updates?
update means something like
sudo dnf update
or something …
apt-get upgrade
apt-get update
Yum-cron. Daily. Rolling bounce on a schedule.
It has been rock-solid for 20 years, but lennart’s cancer and the growing amount of shite they’re shoveling into EL has caused a few issues here and there with 7, 9 and 10. (Skipped 8 because f that)
But, today, it works. So that’s year 23 and 8 months.
Unattended-upgrade does security-only patching once every 4 hours (in rough sync with my local mirror)
Full upgrades are done weekly, accompanied by a reboot
I find that the split between security patching and feature/bug patching maintains a healthy balance knowing when something is likely to break but never being behind on the latest cve.
For me, unattended-upgrade does it’s thing. Updating other packages happens whenever I think about it. Very few things are not containerized and there’s very little added beyond the base Debian install, so when I do update its maybe a dozen packages.
I would previously reboot during thunderstorms if we lost power, but now that I’ve got a UPS I probably ought to come up with a different plan.
All services are dockerized, updated nightly.
Server OS runs a kernel-patch service for real time exploit patching.
All other updates as soon as they appear.
Yeah, sometimes I’ll need to go in a repair - but that’s way better than having to clean up after having been exploited due to not keeping up on security patches.
Those apt commands are in a less-good order. It’s usually better to update apt, then upgrade the system.
I upgrade as soon as reasonably possible after the notification appears, if the system isn’t on auto-upgrade.
I do
sudo apt update && sudo apt upgradeIs there any reason to not combine the commands since the output always prompts prior to changes anyway?
I think their point was to make sure they are done in order, i.e. update before upgrade, not the other way around as in OPs example.
When I remember. About once a month.
Same here. No auto updates, no touching of a stable system without my manual intervention. 😅
Last thing I need in my life is a broken system at home when I don’t have time for it!
All systems, daily via a single ansible script. That’s apt update, upgrade and reboot if needed (some systems set to only reboot with a separate script so I can handle them separately).
Rarely have any sort of problems.
On Alpine Linux I update my two Pi servers at 2 in the morning daily. It’s simpler compared to Debian which needs unattended-updates. Just add
apk update && apk upgradeto a cron job and you’re good to go.I only have three docker services which is simple enough to update manually.
I like to keep things as simple as possible for my already chaotic brain.
Be careful with unattended upgrades, even on alpine. A recent breaking change in python3 broke my alpine 23 ansible instance. Thankfully I have backups, but if you’re going to automate the upgrade, you should automate tests as well.
My web facing server has just enough packages installed to (kinda securely) host a Caddy and Kiwix docker container to work with my domain name and make a comfortable work environment through SSH. My Pi for my HomeAssistant docker container has less because it’s locked down to just my local network.
I also wrote my own install scripts so reinstalling everything and getting it back to a running state would take about 15 minutes for each device.
And I also wrote my own backup/restore scripts that evolved over 3/4 of a year. I use them often so I have confidence in those scripts.
I personally don’t really care too much. I have multiple ways of dealing with issues for something that’s a hobby to me. Which is why I stick to simplicity.
I’m sure this is a thing for people to worry about when dealing with more complex setups. I just wanna vibe out in my tiny corner of the internet.
To make it even simpler,
apk -U upgradeapkseems to have some tricks in there that aren’t as well known.I managed to catch in the IRC channel that
apk add docwill automatically download any related man pages for packages with any future downloads throughapk. That made life a bit more convenient instead of downloading all those packages separately.
Every night at ~ 12-1am
unattended updates / transactional-update are awesome.
Stuff has been running for years, and it’s still up to date.
This is the way! At least install security upgrades nightly using
unattended-upgradesand reboot from time to time to get the latest Kernel version.I wish I could use unattended-upgrade.
It literally restarts my server even when I disable the option, leaving it hung if the USB boot key isn’t in there.
I had to stop using it, so now I just manually upgrade because that doesn’t auto-restart without my permission…
unattended-upgrades doesn’t do that unless you explicitly specify
Unattended-Upgrade::Automatic-Reboot "true";in the config. Check/usr/share/doc/unattended-upgrades/README.md.gzThe main configuration file is
/etc/apt/apt.conf.d/50unattended-upgrades, maybe you put your config in the wrong place?here is mine
This guy scares me
Once per week for me. Works really great on openSUSE MicroOS. Had to roll back maybe a couple of times the last few years.
That said, I run basically everything in containers so the OS installed things are lean.
Automatic upgrades handle the security patches. Everything else maybe once a month. My big services like Nextcloud auto update as well.
I SSH in and run an update manually, once a week.
I’m not knowledgable and comfortable enough to let updates happen automatically and feel like I could trust it to keep running. Not yet, anyway.
Edit: But at some point I might do what another commenter said and make sure security updates run automatically and check other updates weekly.
I do it every 3 to 5 days. I usually do it when I have time to fix things if it goes south.
Got apticron set up on my servers or similar solutions to get notified when updates are available. Then usually, from time of notification +1 or 2 days.
On my ubuntu I use unattended updates but that doesn’t work reliably. I have to update it manually most of the time. Once every other month.
On my fedora server it auto updates every day at 4 reliably.
The next server is going to be atomic such that the server restart is even shorter (not that I would care about it at 4).
everyday to once a month, depending how often I use the server
IME usually waiting longer to apply larger updates causes more issues than smaller and more frequent ones
Every couple of days. I don’t auto-update, but I’ve streamlined the process to the point that I can just open a single web page and see the number of pending updates for every system on my network, docker containers included, each one with a button. Clicking the button applies the update and reboots if necessary. So it takes about 15 seconds of effort to update everything, which is why I don’t mind doing it so often.








