Skip to main content

Tests

In the drive view, the Tests area offers five test types and the test plans. Only one job runs per drive at a time. The durations shown are estimates (about 150 MB/s is assumed for read and write times, and the firmware's time estimate for SMART tests).

TestWhat happensDestructive
SMART short testShort firmware self-test, detects gross errors. Takes a few minutes.no
SMART long testFull surface test by the drive firmware (extended). Takes several hours depending on size.no
Read scanReads the whole drive with badblocks (read only). Data is kept.no
Write testbadblocks writes 4 patterns over the whole drive, reads back and compares. With the "Quick" option only one random pattern is used (about 4 times faster).yes
Burn-inSMART short test, write test and SMART long test one after another.yes

The write test (badblocks -w) destroys all data on the drive.

Safety for destructive tests​

  • Destructive tests only start after explicit confirmation: the serial number (the identifier for drives without a serial number) must be typed in.
  • Drives with mounted partitions, active swap, active RAID/LVM devices or the system/boot drive are blocked for destructive tests. The service enforces this itself, not only the interface. The system drive carries the "System" badge, other occupied drives "In use" with the reason.
  • Drives that only hold old RAID/LVM/ZFS/LUKS signatures are blocked for tests, but can be cleaned with Quick erase (see Erasing).
  • Before each step the service checks again that the same drive is still behind the path, and it refuses to start if another badblocks process is already working on the device.
  • Before every destructive run, check model and serial number in the interface and compare them with the label on the drive.
  • When the service is stopped, running badblocks processes are terminated. Jobs stored as "running" are marked as "interrupted" on the next start.

Test plans​

A test plan is an ordered list of 1 to 8 steps that run one after another as a single job. Each step may appear at most once; the first failed step ends the job. Possible steps: SMART short test, SMART conveyance test, read scan, write test, SMART long test, quick erase, overwrite and verify.

A plan with a writing step (write test, quick erase, overwrite) is destructive: it needs a drive that is not in use, and confirmation by serial number.

Built-in plans (cannot be changed):

PlanSteps
Quick check (quick-check)SMART short test
Full read (full-read)SMART short test, read scan, SMART long test
Burn-in (burn-in)SMART short test, write test, SMART long test
Burn-in quick (burn-in-quick)SMART short test, write test with the "Quick" option, SMART long test
Wipe and check (wipe-and-check)Quick erase, SMART short test

Your own plans are created, changed and deleted under Settings; afterwards they can be chosen in the drive view and for auto mode. Options can be set per plan: "Quick" for the write test, and method (zeros or random), passes and final zero pass for overwriting.

  1. Connect the drive, look at the SMART baseline state and write it down (reallocated, pending, offline uncorrectable, power-on hours, UDMA-CRC errors).
  2. Start the SMART short test, then the SMART long test (extended). Depending on size this takes several hours.
  3. Destructive write test (4 patterns) over the whole drive. On large drives this takes days; progress is visible in the interface.
  4. Afterwards check the SMART values again and compare them with the initial state, then run another SMART long test.

Burn-in combines steps 2 to 4 (short test, write test, long test) into one job. With auto mode a plan starts right after plugging in.

Reading the results​

Each job compares the critical SMART counters (5, 197, 198, 187, 199 and for NVMe the media errors) before and after the run. If a counter rises, a finding such as "Reallocated Sectors: 0 → 8" is created. The drive's test history shows result, duration, findings and log for each job.

The job status is:

  • passed: All steps ran without errors. After each finished step the phase display is reset, so a passed job no longer shows a running self-test.
  • failed / error: A step found errors or failed.
  • aborted: Stopped by the user.
  • interrupted: The service was stopped during the job.

The drive's verdict then rates the SMART state:

  • Good (PASS): no badblocks errors, overall SMART status PASSED, no critical counters grown. The drive can be reused.
  • Warning: minor anomalies (e.g. a few newly reallocated sectors, high power-on hours, CRC errors, often cable or adapter). Keep watching it, do not use it for important data without redundancy.
  • Bad (FAIL): badblocks found errors, a SMART self-test reports read errors, there are pending/uncorrectable sectors or the overall SMART status is FAILED. Discard the drive.

The reason that the verdict lists in the interface is always authoritative. Details on the individual findings are under Usage.