fix: MCP proxy resilience — discovery cache, default liveness probes

Adds a per-server tools/list cache in McpRouter (positive + negative TTL)
so a slow or dead upstream only stalls the first discovery call, not every
subsequent client request. Invalidated on upstream add/remove.

Health probes now apply a default liveness spec (tools/list via the real
production path) to any RUNNING instance without an explicit healthCheck,
so synthetic and real failures converge on the same signal.

Includes supporting updates in mcpd-client, discovery, upstream/mcpd,
seeder, and fulldeploy/release scripts.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

fulldeploy.sh

@@ -1,5 +1,13 @@
 #!/bin/bash
-# Full deployment: Docker image → Portainer stack → RPM build/publish/install
+# Full deployment: mcpd image → k8s rollout → RPM build/publish/install
+#
+# Production runtime is Kubernetes (context: worker0-k8s0, namespace: mcpctl).
+# The docker-compose stack under stack/ + deploy/ is kept for local/VM testing
+# only and is no longer invoked from here.
+#
+# Infra (Deployment shape, env, RBAC, NetworkPolicies) is managed by Pulumi
+# in ../kubernetes-deployment. This script runs `pulumi preview` before the
+# rollout; if there is infra drift it halts so you can `pulumi up` first.
 set -e
 
 SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
@@ -10,19 +18,50 @@ if [ -f .env ]; then
   set -a; source .env; set +a
 fi
 
+KUBE_CONTEXT="${KUBE_CONTEXT:-worker0-k8s0}"
+KUBE_NAMESPACE="${KUBE_NAMESPACE:-mcpctl}"
+KUBE_DEPLOYMENT="${KUBE_DEPLOYMENT:-mcpd}"
+PULUMI_DIR="${PULUMI_DIR:-$SCRIPT_DIR/../kubernetes-deployment}"
+PULUMI_STACK="${PULUMI_STACK:-homelab}"
+
 echo "========================================"
 echo "  mcpctl Full Deploy"
 echo "========================================"
 
+# --- Pre-flight: Pulumi drift check ---
+echo ""
+echo ">>> Pre-flight: checking for Pulumi infra drift"
+echo ""
+if [ -d "$PULUMI_DIR" ]; then
+  if [ -z "$PULUMI_CONFIG_PASSPHRASE" ]; then
+    echo "  WARNING: PULUMI_CONFIG_PASSPHRASE not set — skipping drift check."
+    echo "           Set it in .env or export it to enable."
+  else
+  preview_output=$(cd "$PULUMI_DIR" && pulumi preview --stack "$PULUMI_STACK" --non-interactive --diff 2>&1) || true
+  if echo "$preview_output" | grep -qE '^\s+[-+~]'; then
+    echo "$preview_output"
+    echo ""
+    echo "ERROR: Pulumi detected infra changes that have not been applied."
+    echo "       Run: cd $PULUMI_DIR && pulumi up -s $PULUMI_STACK"
+    echo "       Then re-run this script."
+    exit 1
+  fi
+  echo "  No drift — infra is in sync."
+  fi # passphrase check
+else
+  echo "  WARNING: Pulumi repo not found at $PULUMI_DIR — skipping drift check."
+fi
+
 echo ""
 echo ">>> Step 1/3: Build & push mcpd Docker image"
 echo ""
 bash scripts/build-mcpd.sh "$@"
 
 echo ""
-echo ">>> Step 2/3: Deploy stack to production"
+echo ">>> Step 2/3: Roll out mcpd on k8s ($KUBE_CONTEXT / $KUBE_NAMESPACE)"
 echo ""
-bash deploy.sh
+kubectl --context "$KUBE_CONTEXT" -n "$KUBE_NAMESPACE" rollout restart "deployment/$KUBE_DEPLOYMENT"
+kubectl --context "$KUBE_CONTEXT" -n "$KUBE_NAMESPACE" rollout status "deployment/$KUBE_DEPLOYMENT" --timeout=3m
 
 echo ""
 echo ">>> Step 3/3: Build, publish & install RPM"