Hourly scrape, daily digest, weekly archive: one container

You run a price-watch tool. Every hour you scrape a list of competitor pages with hoody-browser. Every day you roll the data into a digest. Every Sunday you archive the previous week's tables into a file.

That's three cron entries on the same container, each pointing at its own exec script. The hourly job writes raw rows to sqlite. The daily job reads sqlite, computes deltas, writes the digest. The weekly job exports a parquet file to a hoody-files URL and prunes old rows.

You used to need a scheduler service, a job queue, a worker pool, a separate ETL stack. You have one container with three crontab lines, sitting on a flat-rate bare metal server you already pay for. The cadences nest cleanly because they're just three rows of * * * * *.

Three cadences, three cron lines, one container, no scheduler service.