1955294 - OSKeyStore decryption event only submits in failure cases and doesn't include the failure message

Status: RESOLVED FIXED

(Toolkit :: Form Autofill, defect)

Description

[Dave Townsend [:mossop]]

The events submitted from OS keystore decryption added bug 1831049 unfortunately only submit data in failure cases and in those cases do not submit the error message. This is due to two errors:

The first problem is that the property for the error message is defined as error_message but actually submitted as errorMessage. Glean considers pings that include unexpected extra properties as invalid.

However we do still submit in the case where decryption fails because this block causes us to skip setting errorMessage to anything unless the error was the user cancelling the request, Glean does ignore properties with the value undefined so the event submits in this case, but with no error message to tell us what the error was.

The upshot is our current data cannot tell us how often failures are occurring since we can't compare to the number of decryption attempts and we can't see what kinds of errors are occurring.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1831049

:hballent, since you are the author of the regressor, bug 1831049, could you take a look? Also, could you set the severity field?

For more information, please visit BugBot documentation.

Comment 2

[Dave Townsend [:mossop]]

Attachment: Bug 1955294: Fix event ping for OS keystore decryption attempts r=#credential-management-reviewers!

Also changed the error message property to be named errorResult since what we store in here is
the nsresult value not a string message.

Comment 3

[Donal Meehan [:dmeehan]]

It's too late for Fx136.
It's also too late for Fx137 beta since the final Fx137 beta is done and next week is RC week.
:mossop is this something we need to keep on the radar for Fx137, or should it ride the train with Fx138?

Comment 4

[Dave Townsend [:mossop]]

(In reply to Donal Meehan [:dmeehan] from comment #3)

It's too late for Fx136.
It's also too late for Fx137 beta since the final Fx137 beta is done and next week is RC week.
:mossop is this something we need to keep on the radar for Fx137, or should it ride the train with Fx138?

It would be nice to have data from this probe sooner but no there is no real urgency to get it out sooner than 138.

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/fc493856e57d
Fix event ping for OS keystore decryption attempts r=credential-management-reviewers,mtigley

Comment 6

[agoloman]

https://hg.mozilla.org/mozilla-central/rev/fc493856e57d