Parallelize and rebalance bucket C tests

[Amaury Chamayou (achamayou)]

Summary

Follow-up to #8236, which frees substantial capacity in bucket_b and makes bucket_c the end-to-end CI critical path.

This PR:

• splits the virtual code_update_test into independent attestation, node join-policy, and code-upgrade ConcurrentRunner sub-tests;
• keeps consecutive atomic and non-atomic upgrades together, preserving their ordering and coverage;
• keeps the standalone SNP sequence on one network so its final quote verification still covers every policy-join node;
• moves e2e_logging_http2 (~74s) from bucket_c to the newly shortened bucket_b.

The bucket rebalance is intended to land after #8236. Without #8236, bucket_b does not yet have the required headroom.

Expected effect

A recent main run measured code_update_test at 82.4s. Based on its individual phase spans, the parallel virtual test should take approximately 42-47s, saving roughly 35-40 runner-seconds.

After #8236, moving e2e_logging_http2 projects the two virtual jobs at approximately:

Job	Before this PR	After this PR
VMSS Virtual B	~842s	~916s
VMSS Virtual C	~1044s	~930-935s

This PR should therefore reduce end-to-end wall-clock time by approximately 109-114s after #8236. The bucket move only redistributes runner time; the estimated 35-40s runner-time reduction comes from parallelizing code_update_test. These estimates are pending confirmation in CI.

Coverage and testing

• Verified by AST that the virtual code_update_test preserves the exact sequence of all 11 test_* calls.
• Verified by AST that the SNP code_update_test preserves the exact sequence of all 13 test_* calls on one network.
• Re-verified Split schema_test operations into parallel sub-tests #8236: all 29 e2e_operations.run_* calls occur exactly once, all seven groups are registered exactly once, and relative order is preserved within every group.
• Verified that the complete ctest bucket union is unchanged and only e2e_logging_http2 moves from C to B.
• The coverage workflow does not select bucket labels, so the label move does not alter instrumented test coverage.
• uvx black tests/code_update.py
• uvx ruff check tests/code_update.py
• python -m py_compile tests/code_update.py
• Verified runner registration with -N on virtual (attestation, join-policy, code-updates) and SNP (attestation).
• Configured a fresh CMake build and verified the generated bucket inventory matches tests/ci-buckets.txt.
• Full end-to-end timing pending CI.

[Copilot]

Pull request overview

This PR reduces end-to-end CI wall-clock time by parallelising code_update_test into multiple ConcurrentRunner sub-tests (so phases can run concurrently on separate networks), and rebalancing CI buckets by moving e2e_logging_http2 from bucket_c to bucket_b.

Changes:

• Split tests/code_update.py execution into independent ConcurrentRunner sub-tests (attestation, join-policy, code-updates) while preserving ordering where it matters (notably the sequential upgrade coverage).
• Keep the SNP-specific sequence on a single network, including a final test_verify_quotes() after the join-policy-related node activity.
• Rebalance CI by moving e2e_logging_http2 from bucket C to bucket B (both in tests/ci-buckets.txt and the CMake test registration).

Custom instructions used:

• .github/copilot-instructions.md

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
tests/code_update.py	Parallelises code_update_test phases via ConcurrentRunner, separating virtual/non-SNP join-policy and code-upgrade work while keeping SNP flows on a single network.
tests/ci-buckets.txt	Moves e2e_logging_http2 from bucket_c to bucket_b to rebalance CI job durations.
CMakeLists.txt	Updates the e2e_logging_http2 test’s BUCKET assignment to bucket_b to match the intended CI rebalance.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.