Stabilize FlatKV EVM migration stop point

[masih]

The FlatKV EVM migration test could flake while switching validators from memiavl_only to migrate_evm. The script only checked that validators had the same committed height before flipping modes, but a validator can already have signed a vote for the next height without committing it. After the mode flip, consensus WAL replay could ask that validator to sign different data for the same height, triggering Tendermint's double-sign guard with "conflicting data" and leaving the cluster stuck.

Fix the migration choreography by stopping validators, configuring a temporary future halt-height, restarting them in memiavl_only, and waiting for all nodes to halt themselves at the same durable commit boundary. The script also records priv_validator_state heights and refuses to flip if last-sign state advanced past the committed halt height. Reset halt-height when switching to migrate_evm so the migrated restart can make progress.

Flaked for unrelated changes.

[cursor]

PR Summary

Low Risk
Changes are confined to an integration test shell script; no production node or consensus code paths are modified.

Overview
Replaces the pre-flip validator shutdown loop in verify_flatkv_evm_migrate.sh with a halt-height–driven stop so the cluster flips sc-write-mode only after every node halts on the same committed boundary, avoiding Tendermint double-sign flakes when last-sign state is ahead of committed height.

The script stops all validators, sets a temporary halt-height (default +300 blocks from the max of committed height, priv_validator_state last-sign height, and the pre-flip sync floor), restarts in memiavl_only, and waits up to HALT_TIMEOUT. It records last-sign heights and aborts the flip if committed heights diverge or last-sign advanced past the halt commit. halt-height is reset to 0 when switching to migrate_evm.

New helpers factor out coordinated barrier-synchronized starts, parallel TERM stops, and halt-height writes with verification on node 0. The old docker pause / multi-attempt freeze path and PRE_FLIP_STOP_ATTEMPTS are removed.

^{Reviewed by Cursor Bugbot for commit 26ddc6a. 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	Jun 15, 2026, 5:02 PM

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dc65617ab5

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Wait for all validators before starting halt countdown

When one validator restarts slower than the other three, this fixed 10-block window can be consumed before the lagging node is up and caught up: in the local 4-validator devnet any three equal-power validators can form quorum, and the config allows very fast blocks (unsafe-commit-timeout-override = "50ms" in docker/localnode/config/config.toml:358). Those three validators can then halt at HALT_HEIGHT, leaving the fourth without live peers to commit the halt block, so wait_for_all_seid_stop times out instead of stabilizing the migration; choose the halt target only after all validators are observed running/caught up, or make the window large enough for the slowest restart path.

Useful? React with 👍 / 👎.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 58.36%. Comparing base (e7c0eff) to head (26ddc6a).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3592      +/-   ##
==========================================
- Coverage   59.22%   58.36%   -0.87%     
==========================================
  Files        2213     2139      -74     
  Lines      183115   174568    -8547     
==========================================
- Hits       108453   101878    -6575     
+ Misses      64914    63642    -1272     
+ Partials     9748     9048     -700     

Flag	Coverage Δ
sei-db	70.41% <ø> (ø)
sei-db-state-db	?

Flags with carried forward coverage won't be shown. Click here to find out more.
see 74 files with indirect coverage changes

🚀 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.