From 196c9473a4e5ec5411867e1b71bb40c3d019650b Mon Sep 17 00:00:00 2001
From: Graham McIntire <graham@mcintire.me>
Date: Wed, 20 May 2026 12:22:53 -0500
Subject: [PATCH] ops(k8s): point liveness/startup at /health/live, scale to 2
 replicas

Now that the image serving /health/live is live, switch liveness and startup
probes to the shallow endpoint so transient db/redis blips can't kill or block
a healthy pod; readiness stays on the deep /health. Run 2 replicas so a single
pod restart no longer means full downtime.
---
 k8s/deployment.yaml | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/k8s/deployment.yaml b/k8s/deployment.yaml
index b8e9870c..38679e32 100644
--- a/k8s/deployment.yaml
+++ b/k8s/deployment.yaml
@@ -4,7 +4,7 @@ metadata:
   name: towerops
   namespace: towerops
 spec:
-  replicas: 1
+  replicas: 2
   strategy:
     type: RollingUpdate
     rollingUpdate:
@@ -269,9 +269,13 @@ spec:
             limits:
               memory: "3Gi"
               cpu: "2000m"
+          # Startup and liveness use the shallow /health/live (no db/redis) so a
+          # transient dependency outage can't prevent boot or kill a running
+          # pod. Readiness uses the deep /health so the pod is pulled from the
+          # load balancer (but NOT restarted) while db/redis are unreachable.
           startupProbe:
             httpGet:
-              path: /health
+              path: /health/live
               port: 4000
             initialDelaySeconds: 10
             periodSeconds: 5
@@ -279,7 +283,7 @@ spec:
             failureThreshold: 24 # 10s + (24 * 5s) = 130s max startup time
           livenessProbe:
             httpGet:
-              path: /health
+              path: /health/live
               port: 4000
             periodSeconds: 10
             timeoutSeconds: 3