Skip to content

harness: dbx current fetches LVFS's metadata itself - #38

Merged
aledbf merged 1 commit into
mainfrom
fix/dbx-check-refresh
Nov 3, 2026
Merged

aledbf merged 1 commit into
mainfrom
fix/dbx-check-refresh

Conversation

@aledbf

@aledbf aledbf commented Nov 3, 2026

Copy link
Copy Markdown
Collaborator

dbx current, added in #37, failed every task test:usb on the merged tree (enrolled 0). The machine was right and the check was wrong: it waited three minutes for LVFS's metadata, but fwupd's own refresh is an hourly timer with up to an hour's random delay, and GNOME Software's comes when it comes.

The check now runs fwupdmgr refresh --force itself and still requires the refresh to succeed. A cold refresh took 3 s in a VM just after boot.

Checked at this commit

  • task check: passed
  • task test:usb three times: all passed, the last 631/631, with dbx current: enrolled metadata 0 in each
  • Boot: 11.52 s and 12.91 s, against a baseline of 12.97–14.75 s. A third run took 31.91 s, of which the initrd was 12.96 s; the check still passed, and this machine is shared, but the cause was not established.

The check now depends on reaching LVFS: without it, it fails instead of passing on stale data.

🤖 Generated with Claude Code

Waiting for the metadata failed every run of test:usb on the merged tree:
fwupd's own refresh is an hourly timer with an hour's random delay and
GNOME Software's comes when it comes, so three minutes after login the
file was often not there. A cold fwupdmgr refresh takes seconds, and the
check no longer depends on who else got there first.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@aledbf
aledbf merged commit 521d7b6 into main Nov 3, 2026
4 checks passed
@aledbf
aledbf deleted the fix/dbx-check-refresh branch November 3, 2026 14:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant