Deploy to prod

[carlobeltrame]

🏗️ Changes in this deployment

Includes the following non-renovate PRs:

• Rewrite deploy ecamp3 #8025
• chore(api): rector to php80 #8332
• Rector php83 #8418
• chore(api): fix 'Setting PHP_CS_FIXER_IGNORE_ENV environment variable is deprecated' #8432
• docker-compose.yaml: restart api container on-failure #8444
• chore(api): rector to php84 #8431
• Set reply-to on camp invitation mails #8493
• reverse-proxy-nginx: disable dns cache for upstream #8443

Code to generate this overview

The above can be output locally using the following command:

echo 'Includes the following non-renovate PRs:' && git log origin/prod..origin/staging --merges --first-parent --perl-regexp --grep='from ecamp/renovate/' --invert-grep | grep "Merge pull request" | awk '{print "- "$4}' | tac

[carlobeltrame]

/fast-forward

[carlobeltrame]

In this deployment, the domain was wrongly calculated as https://prod.ecamp3.ch. This led to the Kubernetes services not being reachable. When fixing the services to listen to the correct domain, various environment settings such as FRONTEND_URL, API_URL, CORS settings, allowed hosts in the API and so on were still incorrect. When correcting these manually as well, the Cloudflare cache kept on serving the erroneous version of environment.js. After correcting this as well, the services were back up and running.

We should make sure to fix the calculation of the prod domain before deploying the next time. Unfortunately, staging didn't have this problem (because it uses the formulaic subdomain, unlike production).