fix(coprocessor): verify input via eth_getLogs

[dartdart26]

Instead of relying on subscriptions that are not reliable in all cases, use eth_getLogs to fetch verify input requests.

Also, change:

• poll interval from 1 sec to 500 ms to reduce latency

Since gw-listener catches up automatically via the last processed block number in the DB, call the manual process "replay" instead of "catchup". Replay is requested via the cmd line. Also, keep the old name as an alias for backward compatibility. Furthermore, replay is not KMS-specific anymore, so remove KMS from the name.

Add more logging in health checks and replay.

[mergify]

🧪 CI Insights

Here's what we observed from your CI run for b68e72a.

🟢 All jobs passed!

But CI Insights is watching 👀

[Copilot]

Pull request overview

This pull request refactors the gateway listener to use eth_getLogs polling instead of event subscriptions for fetching verify input requests, improving reliability. The terminology changes from "catchup" to "replay" to better reflect the manual recovery process, and the poll interval is reduced from 1 second to 500ms to reduce latency.

Changes:

• Removed subscription-based run_input_verification and merged input verification into the polling-based run_get_logs method
• Renamed catchup_kms_generation_from_block to replay_from_block with backward-compatible CLI alias
• Reduced get_logs_poll_interval default from 1s to 500ms and added configurable logging frequency for last processed block updates

Reviewed changes

Copilot reviewed 6 out of 8 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
gw_listener.rs	Consolidated event processing into run_get_logs, removed subscription-based verification, and added enhanced logging for health checks
lib.rs	Updated configuration field names and defaults for replay functionality and poll interval
bin/gw_listener.rs	Renamed CLI argument with backward-compatible alias and added new logging configuration parameter
gw_listener_tests.rs	Updated test to use new replay_from_block configuration field name
values.yaml	Updated Helm chart with new parameter names and defaults
Chart.yaml	Bumped chart version from 0.7.10 to 0.7.12
Cargo.lock	Updated dependency versions including ruint, cargo-audit, and various windows-sys references
coprocessor-dependency-analysis.yml	Updated cargo-audit version from 0.21.0 to 0.22.0

Comments suppressed due to low confidence (1)

coprocessor/fhevm-engine/gw-listener/src/gw_listener.rs:339

• The VERIFY_PROOF_SUCCESS_COUNTER is incremented twice for each successful verify proof request. It's incremented here in the verify_proof_request function (line 336), and then again at line 276 in run_get_logs via VERIFY_PROOF_SUCCESS_COUNTER.inc_by(verify_proof_success). Remove the counter increment from this function since the caller already handles metric updates after successful DB commit to avoid double-counting.

        inspect(|_| {
            VERIFY_PROOF_SUCCESS_COUNTER.inc();
        }).inspect_err(|_| {
            VERIFY_PROOF_FAIL_COUNTER.inc();
        })?;

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Missing period at the end of the error message. For consistency with other error messages in the codebase (e.g., lines 206, 210, 255), add a period at the end of this error message.

Suggested change

	error!(log = ?log, "Unknown KMSGeneration event")
	error!(log = ?log, "Unknown KMSGeneration event.");

[Eikix]

Already on main, we can close this