Skip to Content
CoreSystemStorage

Storage

System → Storage shows disk usage, flash endurance and write rates, and controls for optional data recording. You can also delete leftover files to reclaim space.

Storage page with a usage bar split into raw sensor archive, app, databases, swap, reclaimable leftovers, and other, and a Reclaimable list with Free up
System → Storage with synthetic disk usage and two leftover items.

What uses space

The bar splits used space into Raw sensor archive, sleepypod app, Databases, Swap, Reclaimable leftovers, and Other. The line under it says how many days of raw sensor history are kept.

The Pod keeps /persistent under a limit (80%, shown by the marker on the bar) by deleting the oldest raw archives. So every leftover file costs you raw history.

Flash endurance

The Flash endurance card shows the eMMC chip’s wear estimate and reserve-block status, plus how much has been written since boot. Wear is reported by the chip in 10% bands; it is not an exact remaining-lifetime forecast. This card is hidden on machines without the Pod’s eMMC counters.

Flash endurance card with one-hour, 24-hour and seven-day daily write rates, measurement coverage, and an hourly bar chart
The real endurance card with synthetic demo measurements.

Last hour, Last 24 hours and Last 7 days show projected GB/day over each window’s measured intervals. The coverage label tells you how many hours were usable; partial means some of that window is missing. Select a bar in the 24-hour chart to inspect its estimated bytes written and coverage.

The rates come from the whole device’s write counter: firmware, sleepypod and OS writes are combined. They are not the amount of new disk space used, and they do not identify an individual service’s write cost. The card cannot yet tell you how many GB/day the Health timeline alone contributes.

At least an hour of usable coverage is needed for a rate. Reboots, counter resets and long sampling gaps are excluded rather than counted as zero writes. Window boundaries are estimated between hourly samples. Write-rate GB uses 1,000,000,000 bytes; the older disk-space figures use binary-sized units.

A deployment or archive job can briefly raise the hourly projection. Compare full 24-hour windows with similar activity before judging an improvement. These counters measure writes sent to the device, not the flash controller’s internal write amplification.

Data recording

Data recording controls showing Quiet app logging, a Debug for 15 minutes button, and enabled Health timeline and raw archive switches
Recording defaults: Quiet logging, with Health history and raw archiving enabled.

Changes persist across restarts. Turning recording off does not delete saved history; existing data remains subject to its normal retention rules.

ControlDestinationWhat changes
App diagnostic logs: QuietSystem journal, configured for RAM by the installerKeeps core console and bundled Python warnings and errors. This is the default.
App diagnostic logs: OffNo new app diagnostic outputSuppresses those app logs, including errors. Firmware, OS logs and direct third-party output are unaffected.
Debug for 15 minutesSystem journalTemporarily permits detailed app logs, then returns to the selected Quiet/Off mode. Stop debug logging ends it early. Expiry survives a restart.
Health timelineFlashStops or resumes new diagnostic timeline samples, taking effect within a minute. Live checks, safety monitoring and module lifecycle status continue.
Raw waveform archiveFlashStops or resumes compressed RAW archives on firmware using the RAM-backed RAW archiver. Live sensing continues; NATS-managed telemetry is unaffected.

Stopping raw archiving asks you to confirm that pending, unsaved waveforms in RAM will be discarded. The linker releases those pins within a minute, while an in-progress archive file may finish. Completed temporary files are still cleaned up so the RAM filesystem cannot fill. Previously saved compressed archives are preserved, subject to normal retention.

Biometric history stays enabled because current vitals and sleep views depend on those records. Hourly flash-write measurements also continue, so you can compare settings.

RAM-backed logs disappear on reboot. Turning logging Off prevents future troubleshooting records from being captured; enabling debug later cannot recover them. Since these logs already use RAM, suppressing them primarily saves processing and memory rather than more flash writes. Smaller log or archive retention limits mainly reclaim space: deleting data does not undo the writes that saved it.

For a bug report, enable temporary debug, reproduce the issue, and open System → Logs to copy or download the relevant output before rebooting.

Free up space

Reclaimable lists files Core left behind, such as a previous release kept after an update or leftover update files.

  1. Review the list.
  2. To also remove older database backups, turn on Include N old database backups. The newest backup of each database is always kept.
  3. Select Free up and confirm with Delete.

Deleted files cannot be recovered. The running app, live databases, raw archive, swap, and HomeKit pairing are never touched.

Over SSH, sp-storage-cleanup --dry-run --json shows the same list without deleting anything. If the page says the cleanup tool is not installed, update Core.


Source reference: Write-rate calculations  · Recording controls  · Recording behavior  · Storage tab  · Storage report  · Operations tools 

Last updated on