Степан Маркелович Печкин рассказывает историю своей семьи

 здесь

тангус и тригометрия на службе у Путина

 "Не секрет, что сейчас используя тангус-синус-косинус QR-тригометрию и лазерную систему наведения криоблокаторов пи-энергии, можно в космосе захватить любую нетяжелую комету (до 30-50 тонн) и изменить ее траекторию."

dybr

 So, this morning I went to prometheus (I don't care about the logs, it's not essential), and saw that the line is just flat for Canary deployment.

Fine! Update the config file to deploy to prod, and wanted to run a build and push - but oops. It does not build. 400 from the server. Bad credentials in a message. (why 400 then?)

What happened? Wrote to our team's slack channel. And then there was a flow of advices. Like "did you log in to vpn?" (did not I?) "did you log in to aws?" (did not I?)

Eventually it turned out that while I was speaking they changed the ids of aws clouds and their names. How come I did not know, right?

Then there was a flow of question to me: - why do I do this, why do I do that?

Guys, I run scripts I wrote half a year ago, after getting information from Safa and Swathi and Karen, now what?

Then Serkan patiently told me what to do, how to update the files; I wrote a PR to update the docs, posted it, and asked Serkan to review. He's a real, true devop guy, and he gave me tons of meaningful advices. That's what I'll be doing any time next time.

And suddenly Santosh wanted to chat on Teams. I would not say I know why, but we did spend about half an hour, and, after many attempts, my files, like `~/.aws/config` etc were properly updated. Yes, the updates were mostly updating. But at least I know what EKS means.

In the end Santosh recommended to write down the commands we ran, in a notepad file (he himself uses VI), so that I don't forget. That's an interesting approach. An Asian approach. Europeans just document stuff.

Then I spent some time updating my deployment config (set the number of instances to 0, and the like). By the end of the day I was ready to deploy to prod (the graph of which was showing how all nodes gobble the memory, but not the canary one which I deployed last night).

And I almost deployed, but strangely, on jenkins, that deployment page was empty, no previous history. So I asked the people. It turned out, ha, when they created a new cloud, the history just disappeared (does it make sense IRL? not for me)>

So, hold on, I'm deploying to prod now.