Emergency production fix from your phone

3:42 AM. PagerDuty. The error rate just spiked.

You don't get out of bed. You unlock your phone, tap the bookmark to your production terminal. You're in: full Linux, real container, rendered in your phone's browser. No app, no VPN, no bastion. The HTTPS URL is the auth.

You tail the log. You see the stack trace. It's the env var change from the 11pm deploy. Typing the fix on a phone keyboard at 3am is hell, so you take the lazy path: you hit the snapshots URL, restore the snapshot from 10:55pm, five minutes before the bad deploy. 28 seconds later production is back.

You message the channel: restored to 10:55pm snapshot, real fix in the morning. You go back to sleep.

At 3:47am, you don't fix bugs. You fix availability. The actual debugging happens at 11am, with coffee, by a human who slept.