Every hour at ~:01–:03 UTC, ALL service deployments in my prod environment are

removed within ~15 seconds — including my Postgres database — while in SUCCESS

state. Nothing replaces them, so the environment stays down until I manually

redeploy, and then it is wiped again at the next hour.

Project: nevo — 6fdbd117-676e-49c1-9454-883c7a0973db

Environment: prod — 28aa9fbb-699c-4edc-adfb-d633e72f7659

Confirmed waves (UTC), from the events API:

Aug 6: 07:01 · 08:01 (9 services) · 09:01 (8 services) · 10:03

Aug 5: 11:03 · 12:02 · 14:03 · 16:02 (8 services each)

After each wave the API domain returns 404 "Application not found" (no container),

and Postgres logs show a clean SIGTERM shutdown — checkpoint completes, then

"Stopping Container". No error precedes it.

Already ruled out:

• Cron schedules: all services now have cronSchedule = null. The Aug 6 10:03

  wave happened with zero crons configured anywhere.

• Usage limits: state = ACTIVE, currentUsage 14.03 of hardLimit 20,

  isOverLimit = false.

• App sleeping: sleepApplication = false.

• My automation: no project tokens, no other members, no webhooks.

• User action: the audit log records my own deploys, but has NO entry for any

  of these removals — they appear only in the events API, with no actor.

• Supersession: no new deployment is created around the waves.

One cron service was oom_killed once (Aug 6, 09:19) and I'm moving that job off

Railway — but the wipes also occur when it isn't running, and my database peaks

at 0.34 GB against an 8 GB limit.

Questions: what is removing these deployments and why is there no audit entry?

Is this platform behaviour, account enforcement, or something in my config?

Would recreating the environment avoid it?