You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Fix reproducibility issues with libyuv and improve tenboxd installation (#117)
* fix: pin libyuv to a specific commit to ensure reproducibility
- Updated the LIBYUV_REF argument in Dockerfile.jammy to a specific commit to avoid issues with NEON inline assembly in the stable branch that breaks the arm64 build-base.
- Modified the git clone command to check out the specified commit, ensuring consistent build outputs across runs.
* fix(linux): reload env file on re-install and clean enabled state on remove
Two reinforcing bugs prevented `install-linux.sh` from reliably
applying /etc/tenbox/tenboxd.env on a re-install:
1. prerm did not call `systemctl disable tenboxd`, so the [Install]
symlink under /etc/systemd/system/multi-user.target.wants/ survived
`apt-get remove`. The next `apt-get install tenbox` therefore saw
is-enabled=true in postinst and ran `deb-systemd-invoke restart`
*before* install-linux.sh had a chance to write the env file,
leaving the daemon running with empty TENBOX_CLOUD_URL /
TENBOX_DATA_DIR and disconnected from the cloud tunnel.
2. install-linux.sh used `systemctl enable --now tenboxd`, which
no-ops the start when the daemon is already active — so even after
the env file landed, systemd never re-read it. Switched to an
explicit `enable + restart` so the freshly written env always wins.
Verified end-to-end: purge -> reinstall -> install-linux.sh now prints
the my.tenbox.ai/pair?code=... URL on the first try, and the daemon's
cmdline carries the correct --cloud-url and --data-dir.
docs/linux-update.md documents the new flow plus the existing
/run/tenbox/tenbox.sock + tenbox-group permission model.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
0 commit comments