From bb87e1a31579e9ae8674f6a468ea954b5d211157 Mon Sep 17 00:00:00 2001
From: Lee Chapman <lee@fullmetal.dev>
Date: Sun, 8 Feb 2026 21:05:10 -0800
Subject: [PATCH] fix: API OOM 137 - liveness probe, memory, probe timeouts

- Use /api/livez for liveness (lightweight, no DB) to avoid SIGKILL on slow health
- Increase memory limit 1Gi -> 2Gi to reduce OOM risk
- Increase probe timeouts: liveness 5s->10s, readiness 3s->10s
- Increase readiness failureThreshold 2->3
- Bump API image to v0.17

Co-authored-by: Cursor <cursoragent@cursor.com>
---
 k8s/api-deployment-blue.yaml  | 14 +++++++-------
 k8s/api-deployment-green.yaml | 14 +++++++-------
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/k8s/api-deployment-blue.yaml b/k8s/api-deployment-blue.yaml
index b873cc1..8fdaa5b 100644
--- a/k8s/api-deployment-blue.yaml
+++ b/k8s/api-deployment-blue.yaml
@@ -20,7 +20,7 @@ spec:
     spec:
       containers:
       - name: api
-        image: harbor.dataknife.net/library/high-command-api:v0.11
+        image: harbor.dataknife.net/library/high-command-api:v0.17
         imagePullPolicy: Always
         env:
         - name: MODE
@@ -56,27 +56,27 @@ spec:
           protocol: TCP
         livenessProbe:
           httpGet:
-            path: /api/health
+            path: /api/livez
             port: http
           initialDelaySeconds: 30
           periodSeconds: 30
-          timeoutSeconds: 5
+          timeoutSeconds: 10
           failureThreshold: 3
         readinessProbe:
           httpGet:
             path: /api/health
             port: http
-          initialDelaySeconds: 10
+          initialDelaySeconds: 15
           periodSeconds: 10
-          timeoutSeconds: 3
-          failureThreshold: 2
+          timeoutSeconds: 10
+          failureThreshold: 3
         resources:
           requests:
             cpu: "200m"
             memory: "256Mi"
           limits:
             cpu: "1000m"
-            memory: "1Gi"
+            memory: "2Gi"
         securityContext:
           runAsNonRoot: true
           runAsUser: 1000
diff --git a/k8s/api-deployment-green.yaml b/k8s/api-deployment-green.yaml
index a80ba1c..f65aa3c 100644
--- a/k8s/api-deployment-green.yaml
+++ b/k8s/api-deployment-green.yaml
@@ -20,7 +20,7 @@ spec:
     spec:
       containers:
       - name: api
-        image: harbor.dataknife.net/library/high-command-api:v0.11
+        image: harbor.dataknife.net/library/high-command-api:v0.17
         imagePullPolicy: Always
         env:
         - name: MODE
@@ -56,27 +56,27 @@ spec:
           protocol: TCP
         livenessProbe:
           httpGet:
-            path: /api/health
+            path: /api/livez
             port: http
           initialDelaySeconds: 30
           periodSeconds: 30
-          timeoutSeconds: 5
+          timeoutSeconds: 10
           failureThreshold: 3
         readinessProbe:
           httpGet:
             path: /api/health
             port: http
-          initialDelaySeconds: 10
+          initialDelaySeconds: 15
           periodSeconds: 10
-          timeoutSeconds: 3
-          failureThreshold: 2
+          timeoutSeconds: 10
+          failureThreshold: 3
         resources:
           requests:
             cpu: "200m"
             memory: "256Mi"
           limits:
             cpu: "1000m"
-            memory: "1Gi"
+            memory: "2Gi"
         securityContext:
           runAsNonRoot: true
           runAsUser: 1000