feat(o11y): execute real inputs through Planner-selected backend plans - #525
Merged
Conversation
This was referenced Sep 8, 2026
zzylol
marked this pull request as ready for review
September 8, 2026 20:15
# Conflicts: # Cargo.lock # control_plane/Cargo.toml # control_plane/src/physical/post_asap/tests.rs # control_plane/src/query_plan.rs # crates/asap_types/Cargo.toml # data_plane/Cargo.toml # data_plane/src/query_engines/asap_query_engine/summary_exec.rs
…nto HEAD # Conflicts: # data_plane/src/precompute_engine/worker.rs
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Run supplied o11ybench inputs through the control plane's normal Planner/cost selection and production data plane, preserving query occurrences, labels and timestamps. Require version-2 complete cost evidence and record actual responses as warm, exact fallback or failed; no winner override or synthetic workload replacement is supplied.
Before this PR, offline binding did not prove that samples were ingested or queries executed. After this PR, the runner installs the selected artifact, sends Remote Write batches, waits for an explicit finite-input completion barrier, and queries every original occurrence. Forwarded responses carry a backend-owned exact-fallback header and cannot count as warm execution.
The completion barrier seals Remote Write before queuing FIFO worker barriers. Workers publish trailing panes before acknowledging. Processing errors and rejected sink outputs remain failures across repeated drains. Valid writes after sealing return HTTP 409; another input generation requires a fresh backend process. This is an explicit finite-replay operation, not a live-ingestion watermark or proof that every closed pane matches every query window.
Validation on the integrated Planner revision abb2f20: 662 control-plane unit tests, 3 Rust drain/receiver regressions, 4 sink tests, 1 fallback-marker regression and 1 production Remote Write wire test passed; 4 Python replay tests passed; git diff --check passed. Integrated #528's counter-safety changes; its process checks remain separate.
Stack: merged #524 and #528 plus deployment/PromQL foundations. #526 adds paired Prometheus comparisons and resource reporting. This PR adds no retained raw-query store, typed logical operator runtime or hybrid execution implementation. A full real-corpus matched run, calibrated costs and reproducible performance/resource benefits are not claimed by these infrastructure tests. Input provenance and raw execution artifacts are preserved for that separate acceptance experiment.