Add EVM stress workload tooling

[codchen]

Summary

• Adds ./scripts/evm_stress.sh, which initializes a local seid node, patches genesis with deterministic sender accounts, starts the node, and runs the transfer workload.
• Adds scripts/evm_stress/main.go, which generates deterministic sender keys, optionally dumps their derived Sei addresses for genesis funding, and sends nonce-0 EVM transfers from unique senders to a shared recipient at the target rate.
• Ignores the local evm_stress build artifact.

This is split out from #3376 so the investigation tooling can be reviewed independently from logging and execution optimizations.

Test Plan

• go test ./scripts/evm_stress
• go run github.com/securego/gosec/v2/cmd/gosec@v2.22.11 ./scripts/evm_stress/...

---

Note

Low Risk
Adds developer-only stress-testing scripts and a Go load generator without modifying production chain logic; risk is limited to local environment usage and potential resource-intensive runs.

Overview
Adds an evm_stress workload toolchain for local performance investigations: a new scripts/evm_stress.sh script initializes a local seid, bulk-patches genesis.json with many deterministic sender accounts, starts the node while filtering key log lines, waits for the EVM JSON-RPC to come up, and then launches a transfer flood.

Introduces scripts/evm_stress/main.go, a deterministic key/address generator that can either dump Sei bech32 addresses for genesis funding or run a rate-limited EVM transfer sender (unique senders, nonce 0) to a single recipient at a target TPS. Updates .gitignore to ignore the local ./evm_stress artifact.

^{Reviewed by Cursor Bugbot for commit 93fda38. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	May 12, 2026, 3:08 AM

[codecov]

Codecov Report

❌ Patch coverage is 0% with 73 lines in your changes missing coverage. Please review.
✅ Project coverage is 59.22%. Comparing base (654d40b) to head (93fda38).

Files with missing lines	Patch %	Lines
scripts/evm_stress/main.go	0.00%	73 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3404      +/-   ##
==========================================
- Coverage   59.25%   59.22%   -0.03%     
==========================================
  Files        2110     2111       +1     
  Lines      174181   174254      +73     
==========================================
  Hits       103210   103210              
- Misses      62044    62117      +73     
  Partials     8927     8927              

Flag	Coverage Δ
sei-chain-pr	0.00% <0.00%> (?)
sei-db	70.41% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
scripts/evm_stress/main.go	0.00% <0.00%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[cursor]

Infinite loop in waitForBalance lacks timeout

Low Severity

waitForBalance spins in an infinite for loop polling for a positive balance, with no timeout or context cancellation check. If the genesis patching fails or the node doesn't produce a block, this hangs forever with no diagnostic output. Since ctx is context.Background(), there's no external cancellation path either.

 

^{Reviewed by Cursor Bugbot for commit 66f51e7. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

There are 2 total unresolved issues (including 1 from previous review).

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

^{Reviewed by Cursor Bugbot for commit 93fda38. Configure here.}

[cursor]

Cleanup fails to kill log-tailing child processes

Medium Severity

kill -- -"$LOG_PID" attempts to send a signal to the process group identified by $LOG_PID, but in a non-interactive shell script (no set -m), background subshells do not become process group leaders — they inherit the parent's PGID. So there is no process group with PGID equal to $LOG_PID, the kill silently fails (masked by 2>/dev/null || true), and the tail/grep children are left orphaned. The comment explicitly states the intent to "kill the entire process group so tail and grep children are also terminated," but this doesn't happen.

 

^{Reviewed by Cursor Bugbot for commit 93fda38. Configure here.}