Repository navigation
feat(dakota): land the Dakota parity base (PPD, presets, defaults, uupd, parity target, firewall/DNS) - #1248
Conversation
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: hanthor The full list of commands accepted by this bot can be found here. Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Danathar
left a comment
There was a problem hiding this comment.
Stand-in review (the hive's reviewers are out of quota). Approving.
Matches Dakota: it gets power-profiles-daemon from gnome-build-meta (core-deps/power-profiles-daemon.bst, listed in Dakota's baselines/dakota/elements.tsv), with no TuneD, tuned-ppd or lpmd, and thermald kept. Dakota's elements/bluefin/powertop.bst leaves powertop's auto-tune off for the same reason.
Checked: nothing still references tuned, intel_lpmd or utah-bootc-kargs-off outside the generated baseline snapshots. The deleted test, profile, tmpfiles entry and post_loaded_profile block go out together. The contract, docs, packages/utah.toml, site data and triage are all updated, and check-doc-counts.py passes at this head. Waiving tuned-ppd is right, because it claims the same D-Bus name as PPD.
Before merge: contract fails with No match for argument: power-profiles-daemon until #1243 (the factory pin) lands, and that in turn needs its nss-mdns contract move. Release note: power modes now use power-profiles-daemon (powerprofilesctl), and custom TuneD profiles and intel_lpmd are gone.
|
@hanthor this PR is back in your court ( Failing checks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (0 waiting): fixes are handed out before new work as soon as you have capacity. |
|
@hanthor this PR is back in your court ( Failing checks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (6 waiting): fixes are handed out before new work as soon as you have capacity. |
a3a1fa9
ℹ️ Sentinel notice — informational; author is trusted (owner), merge is not blockedHive flagged this PR (author @hanthor, head
Hive did not add the |
d39caf3 to
f7934e7
Compare
Dakota (projectbluefin/dakota, gnome-build-meta core-deps) backs GNOME's power-mode switch with power-profiles-daemon and ships neither TuneD, tuned-ppd nor intel-lpmd. Utah now does the same: - install power-profiles-daemon (factory import utah-packages#999) and enable it in the preset, configure-services.sh and the desktop contract. Today tuned-ppd arrives unrequested through gnome-control-center's `Recommends: ppd-service` (it was the only provider); PPD provides and conflicts with ppd-service, so requesting it keeps tuned-ppd out - drop intel-lpmd and its enablement; thermald stays, as on Dakota - remove the TuneD workarounds: the utah-bootc-kargs-off post-loaded profile (PPD has no bootloader plugin, so nothing calls bootc on a profile switch), the /var/log/tuned tmpfiles entry and its test - waive tuned and tuned-ppd in baselines/triage.toml Needs a PACKAGE_IMAGE_SHA bump to a factory publish that carries power-profiles-daemon before the image build resolves it. Assisted-by: Claude Opus via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Dakota (projectbluefin/dakota, testing) is Utah's parity target. Align the system and user presets with what it ships: - sysstat collection is opt-in. Dakota's elements/bluefin/sysstat.bst writes 80-sysstat.preset disabling sysstat.service, sysstat-collect.timer, sysstat-summary.timer and sysstat-rotate.timer. Utah drops the Fedora-derived enables, presets all four off, disables them in configure-services.sh and holds them in the desktop contract's disabled list. The package stays for on-demand sar/iostat. - colord.service and accounts-daemon.service are enabled, as Dakota inherits them from gnome-build-meta's files/systemd-presets/system-preset/80-gnome.preset. - speech-dispatcher.socket and xdg-desktop-portal-rewrite-launchers.service are globally enabled user units, from gnome-build-meta's user-preset 80-speechd.preset and 80-xdg.preset. Preset, configure-services.sh and contracts/bluefin-desktop.toml are wired for every unit, with tests pinning the agreement; the desktop-contract and factory-consumption skill docs follow. Assisted-by: Claude Opus via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Danathar
left a comment
There was a problem hiding this comment.
Stand-in review (the hive's reviewers are out of quota). Approving the consolidated stack. Matching patch IDs against each stacked PR's head: the commits from #1249, #1250, #1251 and #1252 (both of its commits) are identical. The base commit (old #1248), #1253, #1254 and #1273 differ only by rebase fallout: regenerated package counts and generated_at dates, the README's ISO screenshot and CI links, and the nss-mdns doc wording now that the Bluefin baseline is retired. The one new commit moves PACKAGE_IMAGE_SHA to sha256:0b0664cb…, the image whose labels show power-profiles-daemon, uresourced, nss-mdns and the dakota-config/gnome-os-config subpackages built (the problem I raised on #1243). With it, contract now passes, so the stack-wide check-repos no-matches are gone. Note: 0b0664cb was built at 11:48 PM, before #1156 (gdm → pam_oo7) merged, so this pin still has the old gdm. The next pin should wait until #1147 (oo7) is published. The image builds were still pending when I looked.
The factory's colord ships colord.service without an [Install] section (D-Bus activated), so `systemctl enable` is a no-op and the desktop contract's 'required service is not enabled' check failed the nvidia, gaming and nvidia-gaming image builds on #1248. Assisted-by: Claude Opus via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Dakota parity, part 1 (epic #1236). This PR folds the reviewed Dakota stack into one landing so it needs one strict-up-to-date CI cycle instead of nine. Each commit is one of the stacked PRs, unchanged apart from rebases:
Closes #1239, #499 (factory side done).
just checkpasses (1313 tests).Assisted-by: Claude Opus via GitHub Copilot