Infra cleanup, blog pipeline debugging
A day spent documenting pi-infra configs and untangling the auto blog writer pipeline that generates these very logs.
Today was mostly infrastructure housekeeping and fixing my own tooling.
On the pi-infra side, I fixed a real_ip issue on the www vhost so the right client IPs show up in logs instead of Cloudflare’s, and added tracking/documentation for nginx, cloudflared, and fail2ban configs that hadn’t been committed before. Also recorded the removal of an old cron entry (mnl-seo-setup-reminder) that was no longer needed. Separately, I looked into a lighthouse crawl job that wasn’t running on schedule, and reviewed nginx/fail2ban logs for anything unusual — nothing concerning turned up.
The bigger task was debugging the auto blog writer that produces posts like this one. It had stalled out, so I went through staged entries, merged the ones ready to go, and converted a batch of autogenerated “callout” post ideas into straightforward logs instead of full write-ups. I also wired up credentials for the pipeline properly, ran it in dry-run mode to check output, and cleaned up a couple of things in how reports get generated — making sure employer work stays out of them, and moving review notes out of the published post body where they didn’t belong. Fixed the contact email associated with the pipeline output while I was at it.
Quiet otherwise — most of the day was maintenance and fixing the thing that writes these entries.