DevOps: Wpisy

Learn for free 📚 (1/2)

@to3k uwaga techniczna - wpis po „angielsku” - odsyłam do swojego przypiętego wpisu

@74 Fakt to dlatego, że dodany w @ivory, które nie ma jeszcze funkcji zmiany języka :(

Docker apologizes for trying to pull a bait and switch that people pushed back against hard enough for them to reconsider:

Fuck .

Projects need to self-host their container registries. You can get a perfectly cromulent server for a container registry for the money Docker asks for Teams.

Also, I saw suggestions projects move to GitHub Container Registry. "Hey, we just got bitten in the arse by centralization, let's move to another centralized service!" 🤦‍♀️

Also also, support projects financially. They need it. It's free as in freedom, not free as in beer.

Hosting costs. Infrastructure costs. People working on FLOSS projects deserve to a decent wage. Help them help you.

It matters less how much you support them, than how stable that support is. Support them in a way that is comfortable and sustainable for you.

One <local currency> every week or month is better than 20 once. Stability is as important in funding as it is in software!

@rysiek @simon_brooke the problem with decentralised is that it is expensive, being able to self host anything is great in theory, but it always means significant costs in money and, crucially, time. Self hosting anything is almost never a good option, the opportunity cost is too high. The necessary consequence of that is that decentralised services end up morphing into deformed ecosystems dominated by a handful of big providers leveraging economies of scale.

Quite early on they made the decision to stick with a daemon architecture and thus containers effectively ran as root. Understandable from a Windows perspective, but not really state of the art in the Linux world.

@jwildeboer I like the explanation given by one of the founders:

"I was 24 and had no idea what I was doing."

@adamsdesk as a user of images - you will have to be careful, because if the unused namespaces expire there will be a risk of namesquatting and posting malicious images

@miklo Yes, I can understand that.

I was doing some tests on supposed-to-be isolated environments. It turned out this interacted indirectly with production environment. I realized this at that moment I was going to finish work for today. It was 4 PM, right?
Had to upgrade production environment in emergency mode (because it made itself half-upgraded anyway) in the evening, with unfinished configuration and suddenly received big wave of error notifications. It was so quick I almost didn't have time to be really scared. I am not sure what miracle let me fix that.
I didn't even plan to touch production today!

So, how was your ? :blobCat_anxious_sweat:​


:blobcatPats: relatable

@madargon Friday? More like mayday, mayday!

@rafal06 I typically rebuilding the image to fix it automatically.

@chrismckee @maartenballiauw .wslconfig makes the difference in ram/cpu usage ;-) (and yes - it works)

@wikiyu @maartenballiauw I have wsl.conf set; I did try .wslconfig on my laptop, didnt seem to make any difference.
The defaults 50% of total memory on Windows or 8GB, whichever is less; on builds before 20175: 80% of your total memory on Windows

Based on I gave it 12gig of ram/4cores.
Which should restrict the "underlying vm" which isn't a standard vm.
The memory leaks

@wikiyu @grymasTysiaclecia nie no... Wymusiło to restart systemu, bo w tle wpadły aktualizacje ¯⁠\⁠⁠(⁠ツ⁠)⁠⁠/⁠¯

@wikiyu @grymasTysiaclecia uzupełniając wpis nt WSL, trzeba podkreślić ze mowa o WSL2, które znacząco przyśpieszyło deployment, pod warunkiem, ze spelnia sie kilka wymogów. Głownie tworzenie appek wew /home wsl2.

@kravietz Yet another thing to learn and maintain :)


Fun fact: I discovered Dhall while debugging this one 😂


The counterpoint is that if you learn things by seeing how they are done, it teaches you even less of why they are done the way they are.

One thing that, I think, would make getting some of these things easier from how I imagine your POV, is to imagine you have 10x more instances of the thing than you do. This makes it clearer why e.g. making as large as possible part of your system stateless and worrying less about trying to recover it from blunders (instead just making it very easy to reinstall/recreate it from scratch) is something that people often strive for.