Skip to main content

Usage

The web interface is available at http://<host>:3000. It refreshes itself, no reloading is needed. While a test is running, it polls more often.

Dashboard​

The dashboard has two areas and a language switch (DE/EN) in the header:

  • Connected drives: One card per detected drive with model, size, path, label, verdict (traffic light), temperature and the running or last job. Newly connected drives are briefly highlighted. A click opens the drive view. The badge shows "System" only for drives that carry the running system, otherwise "In use: ". If an auto mode countdown is running for the drive, it is shown here and can be cancelled (see Auto mode).
  • Known drives (not connected): All drives the tester has seen before, with their last state. The history is kept until the entry is deleted. Only drives that are currently not connected can be deleted.

System drives (root, /boot, EFI or active swap of the running system) are hidden by default. The Show system disks toggle shows them; the choice is remembered in the browser. All drives, including those not connected, are listed in the Archive.

A notice banner appears at the top when the self-check finds problems (e.g. missing smartctl, no root, no access protection). In mock mode the header says "Mock mode"; the drives are then simulated.

The hotplug poll and the idle SMART refresh can be adjusted under Settings.

Drive view​

The drive view consists of:

  • Identity: Manufacturer (derived from the model), model, serial number, firmware, transport (SATA, USB, NVMe …), size, path and the freely editable fields label and notes.
  • Findings: The verdict with all findings and their reasons.
  • SMART: Key figures (health, temperature, power-on time, power cycles, reallocations, pending and uncorrectable sectors), all attributes, for NVMe the health values, the self-test log and the messages from smartctl. Re-read SMART updates the state immediately.
  • Tests and Erase drive: See Tests and Erasing.
  • Current job: Progress across all steps, estimated remaining time, abort and live log.
  • Test history: All finished jobs of the drive within the retention period (see Archive), each with result, duration and findings.

Tests and erasing are only possible while the drive is connected and no job is running. Destructive actions require the serial number as confirmation.

Label and notes​

Under Identity you can store a short label (e.g. "Bay 3 / old WD Red") and notes. Both belong to the drive (via model and serial number) and are kept after it is unplugged. The label appears on the dashboard and in mails.

Drives without a serial number​

If a USB adapter reports no serial number, the tester tries to identify the drive via SMART. If several drives report the same identity (e.g. adapters with a fixed serial number), each gets its own identifier and destructive actions are blocked for them.

Verdict and findings​

Every finding has a level:

LevelMeaning
InfoNote without effect on the verdict (e.g. high power-on hours, a counter accepted as baseline).
WarningAnomaly that should be watched.
BadSevere finding. The drive should no longer be used for important data.

The verdict is the highest level of all findings: Good (no warning or bad), Warning or Bad. If SMART cannot be read, the verdict is Unknown.

Among other things, the following is checked:

  • Overall SMART status FAILED and attributes below their threshold (currently: bad, previously: warning).
  • Reallocated Sectors (5): bad from 100, warning below. Also reallocation events (196), pending sectors (197), reported uncorrect (187) and spin retry (10) as warning.
  • Offline Uncorrectable (198) and a failed last self-test: bad.
  • Command timeouts (188) above 10 and UDMA-CRC errors (199): warning. CRC errors usually point to the cable, backplane or adapter, not the drive.
  • Power-on hours: above 50,000 info, above 70,000 warning.
  • Temperature above the limit set in the settings.
  • NVMe: critical warning, media errors and insufficient spare capacity are bad, wear of 90 % or more is a warning.

The reason that the respective finding shows in the interface is always authoritative.

SMART baselines​

Counters such as 199 UDMA_CRC_Error_Count (cable/backplane), 5 Reallocated or 197 Pending cannot be reset in the firmware; they count for life. A CRC error that occurred once would remain a warning forever, even after the cable has been replaced.

Accept as baseline (on the respective finding) stores the current value with the date as known. After that:

  • If the value stays the same or drops, the finding is shown only as info and the verdict stays ok.
  • If it rises above the baseline, a warning appears again ("since baseline N on DD.MM.YYYY: +X"), also by mail. A new baseline can be accepted there.
  • Severe findings ("Bad") cannot be accepted. Remove baseline restores the original state.

This tells you whether an error was a legacy case (cable replaced, counter stands still) or continues (counter rises, cable/backplane still suspicious).

Warnings for UDMA-CRC errors, command timeouts, reallocated sectors, reallocation events, reported uncorrect, spin retry and pending sectors can be accepted.

Language​

The interface is available in German and English. The DE/EN switch in the header stores the choice in a cookie (dt_lang); without a choice the language follows the browser, falling back to German. Mails and job logs follow the Language in the settings (default German), since they are not created for a particular viewer.

Settings​

Under Settings you maintain:

  • hotplug poll (seconds) and idle SMART refresh (seconds),
  • language for mails and job logs,
  • retention of jobs in days (30 to 3650, see Archive),
  • your own test plans: create, change and delete,
  • the auto mode,
  • recipients and alert switches, temperature limit and test mail (see Notifications).