ci(ebpf): add retry+timeout to apt-get and docker-build steps - #714
Merged
Conversation
m663 CI experience showed the ebpf-tracing lane's two long-running steps can hang indefinitely on runner network / Docker Hub issues: - `Install eBPF build deps` (apt-get update + install) — stuck for 45+ min during m663 US3 rebase; normally completes in <1 min. - `Container harness (m212 ring_buffer_overflows verification)` (docker build + docker run) — stuck for 45+ min during m663 Polish CI cycle; normally completes in ~10 min. Both are wrapped in `nick-fields/retry@v4` (already used for perf tests). Bounds: - apt-get: 5-min per-attempt timeout, 3 attempts - docker: 20-min per-attempt timeout, 2 attempts Timeouts are generous (5x-10x typical duration) so real slowness still passes; only true hangs get killed and retried. Real failures (kernel-verifier reject, cargo compile error, etc.) bail on the first attempt because they exit non-zero cleanly rather than hanging. ## Not addressed here - Docker Hub outage that prevents the base image from downloading entirely: 2 attempts × 20 min = 40 min max wait, then hard fail. Acceptable — a Docker Hub outage isn't a flake, it's an infra incident. - The underlying "apt/docker occasionally hangs on GHA" pattern is a GHA-side issue; this is the closest we can get without moving off hosted runners. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
m663 CI experience showed the ebpf-tracing lane's two long-running steps can hang indefinitely on runner network / Docker Hub issues:
Both now wrapped in `nick-fields/retry@v4` (already used elsewhere in `perf.yml`). Bounds:
Timeouts are generous (5-10x typical duration) so real slowness still passes; only true hangs get killed and retried.
Real failures aren't retried
Follow-on
Flake fixes done: #1 spdx3-validate retry (#713 merged) + #2 this PR.
Flake #3 (GHA runner image drift) is unfixable from our side — GHA doesn't expose date-pinned runner labels, and the current m234 canary only tests bpf-linker build (not kernel load). Skipping.
🤖 Generated with Claude Code