Repository navigation
Conversation
Danathar
left a comment
There was a problem hiding this comment.
Stand-in review (the hive's reviewers are out of quota). The change itself looks fine. It depends on #664's decision, so I'm commenting rather than approving.
- Remote order: putting
cache.projectbluefin.io:11001beforegbm.gnome.org:11003for bothartifactsandsource-cachesis a clean win on its own. Bluefin-specific artifacts stop paying a GNOME miss first. - Seeding
fsdk-containers.bst:oci/brew-nspawn.bstinkernel-cache.sh: this is only cheap with #664's junction collapse. Without it,brew-nspawnresolves throughfsdk-containers.bst:freedesktop-sdk.bst:…, the separate uncached FSDK graph from #650, so the seed step inbuild.yml(kernel-cache.sh seed release/dev) would try to build that whole graph and could hit the same time limit. So it should land with #664 or after it, and #664 still needs @jeefy's call on the FSDK version skew. If #664 is changed to keep separate junctions, drop the seeding part and keep the remote reorder. - The unit-test update matches the new
ELEMENTS, and the failingzizmoris #651 (pin comments), not this change.
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: castrojo 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 |
|
@castrojo this PR is back in your court ( Failing checks:
The same checks have failed on 2 heads so far (latest 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 (5 waiting): fixes are handed out before new work as soon as you have capacity. |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: castrojo 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 |
Add fsdk-containers.bst:oci/brew-nspawn.bst to ELEMENTS in scripts/kernel-cache.sh so release and dev seeders push the key-free brew machine and its dependency tree to cache.projectbluefin.io:11002. Reorder artifacts and source-caches in project.conf to query the project CAS remote first before upstream mirrors. Assisted-by: GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: castrojo <castrojo@users.noreply.github.com>
2381f7a to
fcb7829
Compare
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: castrojo 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 |
|
/hold Consolidating this cache-seeding fix into #664 so the shared SDK graph, cache population, and required action-pin correction land together, before dependencies. No separate full build needed for this branch. |
Problem
freedesktop-sdk.bst:components/linux.bstandcomponents/go.bstare pushed tocache.projectbluefin.io:11002byscripts/kernel-cache.sh, the Homebrew machine image (fsdk-containers.bst:oci/brew-nspawn.bst) and its dependency tree (brew-deps.bst,brew-runtime.bst) were not seeded. When cold runners buildhomebrew/image.bst, the brew container layers are missing from the CAS.project.conf,https://gbm.gnome.org:11003was listed beforehttps://cache.projectbluefin.io:11001, forcing round-trip misses against GNOME CAS on all Bluefin-specific artifacts and custom kernels before falling back to the project cache.Solution
fsdk-containers.bst:oci/brew-nspawn.bsttoELEMENTSinscripts/kernel-cache.sh(verified key-free: does not reachbluefin-server/keys/boot-keys.bst).https://cache.projectbluefin.io:11001as the first remote inproject.conffor bothartifactsandsource-caches.tests/unit/test_kernel_cache.pyto verify the expandedELEMENTSlist and skip behavior.Critical Sequencing Note
PR #664 MUST merge before this PR.
buildwaits onkernel-cache. If #665 were to land before #664, the release seeder would compilebrew-nspawn.bstagainst the un-collapsed duplicate FSDK graph (359 elements from scratch), causing multi-hour delays and producing cache keys that #664 immediately obsoletes. Once #664 is onmain, #665 seedsbrew-nspawn.bstagainst the host SDK where base elements are already warm.Verification
just validatepasses (799 subtasks resolved).pytest tests/unit/test_kernel_cache.py tests/unit/test_ci_workflows.pypassed (45/45).