Could not run the fingerprint test

The self-test opens a throwaway copy of the browser carrying this profile exact seed and asks it questions. This line means the copy never got far enough to answer, so you are looking at no score rather than at a bad one.

No score beats a wrong score

The test measures the engine your profiles actually launch with, and it prepares the same protection a real profile gets.

Without that protection the answers would show failures your profile does not have, so the app declines to measure instead of showing them.

The test never touches your profile folder

It runs in a temporary directory carrying the same seed, so an open session is never disturbed and nothing is written into the profile.

The engine did not start

The measurement runs the same engine a launch runs, without a window. If the engine is missing or cannot start there is nothing to measure.

A separate line names that case directly, so the short version here usually means something else.

No room on the disk

Preparing the protection for the throwaway copy needs free space, and a full disk stops it. The line for that case says so and asks you to check free space.

It waited and gave up

The measurement gives the browser twenty five seconds to come up and load a page. A machine busy opening a dozen profiles can spend the whole budget.

Close the other profiles and run the test again on a quiet machine.

A profile from a newer build

A profile written by a newer version of the app is not measured, because its seed may carry fields this build does not know. Updating the app brings it back.

Collecting the evidence

Close the other profiles, run the test again so the machine is free, and copy the whole line from the panel: the longer version names the cause when there is one.

Write from the Support screen with the profile name. The report that travels with the message names the engine and its version, and the seed itself is cut out.

What this article does not cover

A low score. That is a measurement that ran, and every failing row names what it checked.
Fingerprint sites elsewhere. They ask different questions and score them their own way.
A profile that will not launch. The test is read only and says nothing about launching.

Where this is decided

Each claim above is one place in the code and one stand.

The audit rules are one source shared by the page and the stand: apps/desktop/test/fp-audit.js
The audit really runs in a real browser rather than being read as text: apps/desktop/test/audit-on-builtin.js
What the engine does not close the page closes so the score is honest: apps/desktop/test/дыры-движка-закрыты.js
Measurement directories are swept even after a run that was killed: apps/desktop/test/probe-dirs-swept.js

Questions

Does the test open my profile?: No. It opens a separate copy in a temporary folder carrying the same seed, so a session you have open is left alone.
Why does it need the protection turned on?: Because a real profile has it. Measuring without it would show failures that do not exist and send you fixing something that is not broken.
Does it leave files behind?: The folder is removed at the end, and leftovers from a run that was killed are swept on the next start.

Next to this

Still stuck

Support reads the report that travels with your message and answers with the cause, not with a checklist.

Contact support Knowledge base