OpenClaw Signals: Cluster Failover Fixes, Transport Bumps, and Memory Guards

PR

PR #333: Channel State Memory Safety Hardening

This PR tightens the internal memory model for channel state management, reducing silent data corruption in high-concurrency environments. It's a critical stability patch if you're running concurrent workloads against shared state.

PR #332: Config Key Standardization Cleanup

Standardizing the configuration keys in PR #332 nukes a class of hardcoded typos that silently break deployments. It's boilerplate cleanup, but it saves you hours of debugging why your overrides aren't applying.

PR #331: Transport Layer JSON Native Parsing

PR #331 shifts the internal transport layer to rely more on native JSON parsing, which should give you a modest latency bump on the wire. It's a low-level refactor, but it's the kind of change that makes single-node setups feel snappier.

PR #328: GC Loop Guardrails for Metric Counters

In #328, the team adds explicit guardrails around the garbage collection cycle for temporary metric counters. This stops memory creep during long-running analysis tasks, which is a silent killer for background services.

Issue

Issue #1564: Failover Race Condition Under Load

Digging into issue #1564 reveals a gnarly race condition in the secondary node failover logic. This is essential reading for anyone operating clustered deployments, as a fix here prevents split-brain scenarios during network partitions.

Issue #1563: CLI Initialization Deadlock Reporting

Issue #1563 flags a potential deadlock in the CLI initialization sequence when local caches are missing. If your scripts hang inexplicably, this thread explains the dependency loop.