starknet_transaction_prover: origin-level breadcrumbs for validation errors

[avi-starkware]

Splits validation-error logging so each origin (request body, RPC,
proving) carries an event tag and trimmed payload context, making it
straightforward to root-cause production failures from the structured
log stream. Also adds an in-test microbenchmark for HttpMetricsLayer
overhead so any regression in the hot path is visible.

Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com

[cursor]

PR Summary

Medium Risk
Observability-only changes to production logs; misclassifying an error variant could leak transaction data or omit useful detail, but proving behavior is unchanged.

Overview
Proving failures that leave the service no longer log full Display strings when they might include client transaction data (fees, hashes, revert reasons, node URLs, program output). A new may_embed_transaction_data on VirtualSnosProverError treats every variant as sensitive except TransactionBlocked, with tests for reverts and payload-carrying validation errors.

Origin-level warn! breadcrumbs in the virtual SNOS prover now use fixed event / reason codes (e.g. validation_error, os_run_error, proving_error) instead of echoing the error. The RPC handler logs the bounded outcome metric label on every failure and only attaches error when the variant is safe to print verbatim.

Background proving during the blocking-check path is **instrument**ed with the current span so request context still flows into the spawned task.

^{Reviewed by Cursor Bugbot for commit f9f5e5d. Bugbot is set up for automated code reviews on this repo. Configure here.}

[avi-starkware]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• starknet_transaction_prover: log resolved config at startup (secrets redacted) #14684
• starknet_transaction_prover: origin-level breadcrumbs for validation errors #14172 👈 (View in Graphite)
• starknet_transaction_prover: /health returns 503 when service is saturated #14171
• starknet_transaction_prover: track saturation and queue depth on the request path #15027
• starknet_transaction_prover: bump prover_panics_total on panic hook fire #14170
• starknet_transaction_prover: HTTP request count + latency + in-flight metrics #14169
• starknet_transaction_prover: proving-job duration + outcome metrics #14168
• starknet_transaction_prover: Prometheus /metrics endpoint with build_info #14167
• starknet_transaction_prover: extract the TLS accept loop and test its drain #15049
• starknet_transaction_prover: graceful shutdown on SIGTERM/SIGINT #15026
• starknet_transaction_prover: global panic hook with structured log #14166
• starknet_transaction_prover: document existing logging, request ids and health probe in README #14203
• main-v0.14.4

This stack of pull requests is managed by Graphite. Learn more about stacking.

[reviewable-StarkWare]

This change is 

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 32a07a2. Configure here.}

[cursor]

Orphan breadcrumbs on blocked path

Medium Severity

Origin breadcrumbs in run_and_prove emit as soon as the parallel prove task fails, but prove_with_blocking_check can still discard that result and return TransactionBlocked. With the default multi-second check timeout, a fast OS or proving failure often finishes first, so the same request span can show os_run_error or proving_error alongside a final failure_blocked outcome and mislead root-cause analysis.

Additional Locations (2)

• crates/starknet_transaction_prover/src/proving/virtual_snos_prover.rs#L254-L259
• crates/starknet_transaction_prover/src/proving/virtual_snos_prover.rs#L324-L327

 

^{Reviewed by Cursor Bugbot for commit 32a07a2. Configure here.}

[cursor]

Output parse mislabeled as proving

Low Severity

The proving_error breadcrumb wraps all of prove_virtual_snos_run, including try_into_proof_facts failures that become ProgramOutputError. Those map to failure_output_parse in metric_outcome, so the origin event and the final outcome disagree and point operators at the wrong stage.

 

^{Reviewed by Cursor Bugbot for commit 32a07a2. Configure here.}