Phase 654: runMeshResilienceSoak journey with CI skip path.

3-peer live soak or single-host heal + replication lag assert when PEARCORD_SKIP_MESH_ROUNDTRIP=1.

Co-authored-by: Cursor <[email protected]>

headless.js

@@ -1942,6 +1942,11 @@ class HeadlessSession {
     return runFanout(scenarioPath, opts)
   }
 
+  async runMeshResilienceSoak (scenarioPath, opts = {}) {
+    const { runMeshResilienceSoak: runSoak } = require('./mesh')
+    return runSoak(scenarioPath, opts)
+  }
+
   /** Journey: simulate remote hub post ingest on discovery mesh. */
   async runFederationHubIngestJourney () {
     if (!this.platform) throw new Error('headless not started')

mesh.js

@@ -315,6 +315,62 @@ async function runMeshChurnScaleSoakJourney (scenarioPath, opts = {}) {
 /**
  * Production federation live: 3-peer mesh + 8 fanout message tokens (CI-safe).
  */
+/**
+ * Phase 654: partition heal + replication diagnostics lag assert.
+ * Full 3-peer live mesh unless `PEARCORD_SKIP_MESH_ROUNDTRIP=1`.
+ */
+async function runMeshResilienceSoak (scenarioPath, opts = {}) {
+  const maxLagMs = Math.max(0, Number(opts.maxReplicationLagMs) || 120000)
+  if (process.env.PEARCORD_SKIP_MESH_ROUNDTRIP === '1') {
+    const base = path.join(os.tmpdir(), `pearcord-agentctl-resilience-${Date.now()}`)
+    const host = new HeadlessSession({ storagePath: path.join(base, 'host') })
+    if (opts.sessions) opts.sessions.push(host)
+    await host.start()
+    try {
+      if (scenarioPath && fs.existsSync(scenarioPath)) {
+        await host.runScenario(scenarioPath)
+      } else {
+        await host.registerUser({ username: 'resilience_host', displayName: 'Resilience' })
+        await host.createGuild({ name: 'Mesh Resilience' })
+      }
+      const gid = host.platform.guild?.guild?.id
+      if (!gid) throw new Error('mesh resilience: no guild')
+      const heal = await host.runPartitionHealJourney(gid)
+      const meshDiag = host.platform.exportMeshReplicationDiagnostics(gid)
+      const lag = Number(meshDiag?.maxReplicationLagMs) || 0
+      if (lag > maxLagMs) {
+        throw new Error(`mesh resilience: replication lag ${lag}ms > ${maxLagMs}ms`)
+      }
+      return {
+        peerCount: 1,
+        guildId: gid,
+        heal,
+        meshDiag,
+        soak: heal?.heal?.throttled ? 'heal-throttled' : 'ok',
+        maxReplicationLagMs: lag,
+        gossipLanes: meshDiag?.gossipLaneCounts || null,
+        ciSkipMesh: true
+      }
+    } finally {
+      await host.close()
+    }
+  }
+  const live = await runPartitionHeal3PeerLiveJourney(scenarioPath, opts)
+  const lag = Number(live.meshDiag?.maxReplicationLagMs) || 0
+  if (lag > maxLagMs) {
+    throw new Error(`mesh resilience: replication lag ${lag}ms > ${maxLagMs}ms`)
+  }
+  if (!live.heal?.heal && live.heal?.throttled) {
+    return { ...live, soak: 'heal-throttled' }
+  }
+  return {
+    ...live,
+    soak: 'ok',
+    maxReplicationLagMs: lag,
+    gossipLanes: live.meshDiag?.gossipLaneCounts || null
+  }
+}
+
 async function runMesh8PeerFanoutJourney (scenarioPath, opts = {}) {
   const triple = await connectGuildMeshTripleGroup(scenarioPath, {
     ...opts,
@@ -339,5 +395,6 @@ module.exports = {
   connectGuildMeshTripleGroup,
   runPartitionHeal3PeerLiveJourney,
   runMeshChurnScaleSoakJourney,
+  runMeshResilienceSoak,
   runMesh8PeerFanoutJourney
 }