On multiple occasions I've listened to instance admins speak about high S3 costs. The sheer amount of data absolutely balloons the more activity your server sees, I get it.
What I don't get is whether there's some unknown fedi ethical reason everybody insists on setting up an S3 cache (followed immediately by complaining about it).
Y'all want to know what the rest of the web does? Hosts their own uploaded media, and links out to the rest...
I've just finished migrating toot.community to Hetzner. It's now operating on Talos Linux with a managed Kubernetes setup across several bare-metal servers.
Utilizing managed Kubernetes on VMs proved to be increasingly costly. While DigitalOcean's managed experience is great, a shift in strategy is needed to maintain financial stability due to the gradual decline in donations each month.
As always, it's a financial trade-off between being cared for, and taking care of.
In a customer session yesterday there was a misunderstanding about "long-lived teams". People felt that the need to keep teams around long-term was a high bar preventing swarming and discovery.
"Long-lived teams" is instead of short-lived teams (2 weeks, 3 months, etc.) that ignore the "superpowers" of well-functioning teams.
Keep the teams together after they have developed a good working pattern, then bring work to that team.
What do you like to read for work or in your spare time? (That's related to your job or code more generally). Trying to get a list of content that's actually good and read by real people vs propped up by algorithms.
Send links if possible or give names of sites / blogs / creators.
Is it a #Linux distro? Or is it a #container orchestration tool?
At my last job our #webDev env was managed by it, but I was using it on #MacOS and we also had to have #Docker, so I could honestly never figure out what it was there for.
A surprisingly common mistake people do when contributing to #OpenSource projects is to forget the (often required) sign-off on their commit, and then close the PR only to open a new one where the sign-off is included. This isn’t needed! Next time, just:
git commit --amend --signoff
git push --force
And your signoff will be added to the commit in your PR.
I know this is a hot topic and I don't want to be guilty of bike shedding, but boy do I wish the #fedora#centos#redhat installer would not create a separate home partition when I select automated partitioning. #Debian asks, rh just assumes this is 'correct'. Why? What advantages does this have, really? #linux#sysadmin#devops
My take on #DevOps these days is that there are no clever and elegant solutions out there for small scale infrastructure needs, even if those are the majority of the backend systems out there.
So many things are invented for large-scale and high demand services and then supposedly tickle down to smaller systems.
The thing is that I don't need a real-time job scheduler for my Rails app image processing pipeline. I really just need a cron job that runs every 10 minutes or so during office hours.
So I have a Wordpress website which is deployed to the apache+nginx server with Ansible. A symlink in the webapp's document root is updated to point to the new release.
I'm encountering Apache (I think) caching the results and serving the index.php page in the old release until I modify the file. Even though the symlink points to the new release.
Any suggestions as to how to fix this other than tweaking the modified date on the old release?
Does anyone know of a tool to monitor #redis pub/sub topics? e.g., messages per second on topics, average message size in bytes, etc.
I know the INFO command gives total stats for all commands, but it'd be nice to have more detail into pub/sub. I'm sure we used to have this at pusher.com, but not sure how, and that was over a decade ago, before they moved to an in-house developed messaging infrastructure
My Python script that scans /var/log/mail.log for abnormalities failed in Debian 12. Upon reading the documentation, I come across this from Debian Wiki 🤡 Systemd is truly pain in the butt. Now I have choices here: A) update my script to use systemd journals to scan logs. B) Install rsyslogd and reconfigure old behavior. Three is 3rd choice go back to FreeBSD or Alpine. But that is too much work. LOL. #debian#linux#devops#sysadmin#opensource
If you think being 100% correct and precise in your instruction manuals will guarantee people will not fuck up using your system, sit down with someone reading your manual sometime to be cured of that notion.
First thing I noticed is that people don't read manuals like novels. They will open your manual in the middle, browse for things that stand out as something they know already, and do that first, regardless of your manual listing 40 steps that need to be done beforehand.