Auto Vacuum handles the common case: your instance sleeps, we compact data.db and logs.db overnight, you wake up with smaller files. No button required.
But what if you just deleted a million log rows and want the disk back now, while the instance is still warm?
That is what Vacuum Now is for.
Auto Vacuum vs Vacuum Now
| Auto Vacuum | Vacuum Now | |
|---|---|---|
| Trigger | Nightly PM2 job | You, from the dashboard |
| When it runs | Instance is idle (hibernated) | Any time you ask |
| Downtime | Up to ~5s if traffic collides | Brief, expected (we stop the container first) |
| Default | On for every instance | Opt-in per action |
Auto Vacuum skips running containers on purpose. Vacuum Now does the opposite: it force-stops the instance, compacts both SQLite files, then lets you power back on. Plan for a short maintenance window.
Why we are building it
Some of you run heavy log retention or bulk deletes during the day. Waiting for hibernate is fine for steady-state bloat. It is not fine when you are debugging disk usage at 2 PM and logs.db is still 4 GB after the admin UI says “500 rows.”
Vacuum Now reuses the same health compact machinery and edge vacuum locks as the nightly sweep. The difference is intent: you asked, we drain in-flight requests, take the lock, compact, report bytes reclaimed, and hand control back.
What to expect in the dashboard
We are adding a Vacuum Now action on the instance page (near Auto Vacuum in Danger Zone). Click it when you want an immediate compact. The UI will show progress and how much space came back.
If the edge cannot grant a lock or disk budget is too tight, you get a clear error instead of a silent skip.
Already self-hosting?
You can compact a specific idle instance today with pockethost health compact. Vacuum Now is the hosted UX for “I need this while the instance is running” without SSHing in.
We will announce here when the dashboard button is live.
Questions? Discord.