Problem / motivation
projectbluefin/testsuite#501 proposes distrobox container-create / package-install / app-export integration scenarios (tests/dx/features/dx_tools.feature, tagged @requires_cached_image). They're implemented and merged (testsuite PR #720) but stay @pending (skipped by tests/shared/quarantine.py) because the test VM has no locally-cached registry.fedoraproject.org/fedora-toolbox:latest image — pulling it fresh inside the e2e run would be slow and flaky over the QEMU SLIRP network, the same class of problem bluefin's e2e.yml already avoids for brew-setup.service/flatpak-preinstall.service by masking them instead of paying for a fresh network pull every run (see projectbluefin/testsuite#487).
I checked projectbluefin/lab for an existing tracker (searched "distrobox", "image cache", "toolbox") and found none — this issue is that tracker, so the pending scenarios have somewhere to link once it's resolved.
Proposed solution
Provision registry.fedoraproject.org/fedora-toolbox:latest into whatever image cache mechanism the test VM already uses for its other prepulled images (base OS images, akmods, etc.), so a podman pull/distrobox create inside the e2e run hits a local cache instead of the network.
Files to change
Unknown from the testsuite side — the caching mechanism lives in this repo (projectbluefin/lab), not projectbluefin/testsuite. Whoever picks this up will know where the existing image-cache provisioning lives.
Acceptance criteria
🐝 Hive Agent: contributor | SHA: a01bac34
Problem / motivation
projectbluefin/testsuite#501proposes distrobox container-create / package-install / app-export integration scenarios (tests/dx/features/dx_tools.feature, tagged@requires_cached_image). They're implemented and merged (testsuite PR #720) but stay@pending(skipped bytests/shared/quarantine.py) because the test VM has no locally-cachedregistry.fedoraproject.org/fedora-toolbox:latestimage — pulling it fresh inside the e2e run would be slow and flaky over the QEMU SLIRP network, the same class of problembluefin'se2e.ymlalready avoids forbrew-setup.service/flatpak-preinstall.serviceby masking them instead of paying for a fresh network pull every run (seeprojectbluefin/testsuite#487).I checked
projectbluefin/labfor an existing tracker (searched "distrobox", "image cache", "toolbox") and found none — this issue is that tracker, so the pending scenarios have somewhere to link once it's resolved.Proposed solution
Provision
registry.fedoraproject.org/fedora-toolbox:latestinto whatever image cache mechanism the test VM already uses for its other prepulled images (base OS images, akmods, etc.), so apodman pull/distrobox createinside the e2e run hits a local cache instead of the network.Files to change
Unknown from the testsuite side — the caching mechanism lives in this repo (
projectbluefin/lab), notprojectbluefin/testsuite. Whoever picks this up will know where the existing image-cache provisioning lives.Acceptance criteria
registry.fedoraproject.org/fedora-toolbox:latestis available locally on the e2e test VM without a network pull@requires_cached_imagescenarios intestsuite'stests/dx/features/dx_tools.featurecan be un-pended (remove@pending) and pass reliably🐝 Hive Agent:
contributor| SHA:a01bac34