fix: repair the Fly console image build - #1241
Merged
Merged
Conversation
Contributor
📝 WalkthroughWalkthroughThe Fly deployment job now limits execution to 60 minutes and disables persisted checkout credentials. The Fly builder image now includes additional tokenizer, serving plugin, host, and script library sources. ChangesFly deployment
Estimated code review effort: 1 (Trivial) | ~5 minutes Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
The Fly console image could not be built, so the Deploy Fly Console workflow failed for any ref. Two build-context gaps in fly/Dockerfile: - scripts/build-llama.sh sources scripts/lib/cuda-toolkit.sh, but only prepare-llama.sh and build-llama.sh were copied into the image, so the llama.cpp build step aborted with 'No such file or directory'. - The hand-maintained crate copy list had gone stale and was missing skippy-tokenizer, mesh-native-serving-plugin-api, and mesh-native-serving-plugin-host, so 'cargo build --locked' could not load the workspace manifest.
michaelneale
force-pushed
the
fix/fly-dockerfile-build
branch
from
August 11, 2026 08:31
0f59bd0 to
4e768e3
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The Deploy Fly Console workflow currently fails for every ref, because the Fly console image cannot be built at all. This restores it, so deploying
public.meshllm.cloudworks from the Actions tab again.Both causes are build-context gaps in
fly/Dockerfile:Missing
scripts/lib/.scripts/build-llama.shsourcesscripts/lib/cuda-toolkit.sh, but the image only copiedprepare-llama.shandbuild-llama.sh. The llama.cpp step aborted after applying the whole patch queue:Stale crate copy list.
fly/Dockerfileenumerates every workspace crate by hand, and had fallen behind by three:skippy-tokenizer,mesh-native-serving-plugin-api,mesh-native-serving-plugin-host.cargo build --lockedthen failed to load the workspace manifest:Validation
Both fixes were applied and used to deploy the real app before this PR was opened —
mesh-llm-consoleis now on v0.75.1 (up from 0.72.1), release v94:Verified after deploy:
/v1/models, non-streaming inference, SSE streaming,model=meshMoA routing, and the console UI.Note
The hand-maintained crate list in
fly/Dockerfilewill keep going stale — nothing verifies it, anddocker.ymlonly runsbuildx build --check(a linter, not a build) on manual dispatch.docker/Dockerfile.clienthas the same two problems today and is missing 14 crates. Replacing the enumeration withCOPY crates/would fix this class of bug permanently, since.dockerignorealready excludestarget/,node_modules/, anddist/. Left out of here to keep this a minimal repair.