cleanup crawl config, add separate 'runNow' field

crawler: add cpu/memory limits
minio: auto-create bucket for local minio

backend/crawls.py

@@ -73,8 +73,9 @@ class CrawlConfigIn(BaseModel):
     """CrawlConfig input model, submitted via API"""
 
     schedule: Optional[str] = ""
+    runNow: Optional[bool] = False
 
-    storageName: Optional[str] = "default"
+    #storageName: Optional[str] = "default"
 
     config: RawCrawlConfig
 
@@ -84,8 +85,9 @@ class CrawlConfig(BaseMongoModel):
     """Schedulable config"""
 
     schedule: Optional[str] = ""
+    runNow: Optional[bool] = False
 
-    storageName: Optional[str] = "default"
+    #storageName: Optional[str] = "default"
 
     archive: Optional[str]
 

backend/dockerman.py

@@ -1,3 +1,5 @@
+# pylint: skip-file
+
 from archives import Archive
 from crawls import CrawlConfig
 

backend/k8sman.py

@@ -7,9 +7,6 @@ import json
 
 from kubernetes_asyncio import client, config
 
-# from fastapi.templating import Jinja2Templates
-from jinja2 import Environment, FileSystemLoader
-
 
 # ============================================================================
 DEFAULT_NAMESPACE = os.environ.get("CRAWLER_NAMESPACE") or "crawlers"
@@ -31,11 +28,6 @@ class K8SManager:
 
         self.namespace = namespace
 
-        loader = FileSystemLoader("templates")
-        self.jinja_env = Environment(
-            loader=loader, autoescape=False, lstrip_blocks=False, trim_blocks=False
-        )
-
         self.crawler_image = os.environ.get("CRAWLER_IMAGE")
         self.crawler_image_pull_policy = "IfNotPresent"
 
@@ -97,15 +89,17 @@ class K8SManager:
         )
 
         # Create Cron Job
-        run_now = False
-        schedule = crawlconfig.schedule
         suspend = False
-        if not schedule or schedule == "now":
+        schedule = crawlconfig.schedule
-            if schedule == "now":
+
-                run_now = True
+        if not schedule:
             schedule = DEFAULT_NO_SCHEDULE
             suspend = True
 
+        run_now = False
+        if crawlconfig.runNow:
+            run_now = True
+
         job_template = self._get_job_template(cid, labels, extra_crawl_params)
 
         spec = client.V1beta1CronJobSpec(
@@ -205,6 +199,23 @@ class K8SManager:
         if extra_crawl_params:
             command += extra_crawl_params
 
+        requests_memory = "256M"
+        limit_memory = "1G"
+
+        requests_cpu = "120m"
+        limit_cpu = "1000m"
+
+        resources = {
+            "limits": {
+                "cpu": limit_cpu,
+                "memory": limit_memory,
+            },
+            "requests": {
+                "cpu": requests_cpu,
+                "memory": requests_memory,
+            },
+        }
+
         return {
             "spec": {
                 "template": {
@@ -227,6 +238,7 @@ class K8SManager:
                                 "envFrom": [
                                     {"secretRef": {"name": f"crawl-secret-{uid}"}}
                                 ],
+                                "resources": resources
                             }
                         ],
                         "volumes": [

backend/requirements.txt

@@ -1,6 +1,5 @@
 uvicorn
 fastapi-users[mongodb]==6.0.0
 loguru
-jinja2
 aiofiles
 kubernetes-asyncio

chart/templates/main.yaml

@@ -26,6 +26,8 @@ stringData:
 {{- if .Values.minio_local }}
   MINIO_ROOT_USER: "{{ .Values.storage.access_key }}"
   MINIO_ROOT_PASSWORD: "{{ .Values.storage.secret_key }}"
+
+  MC_HOST: "{{ .Values.minio_scheme }}://{{ .Values.storage.access_key }}:{{ .Values.storage.secret_key }}@{{ .Values.minio_host }}"
 {{- end }}
  
   STORE_ACCESS_KEY: "{{ .Values.storage.access_key }}"
@@ -52,6 +54,23 @@ spec:
         app: {{ .Values.name }}
 
     spec:
+
+{{- if .Values.minio_local }}
+      initContainers:
+        - name: init-bucket
+          image: {{ .Values.minio_mc_image }}
+          imagePullPolicy: {{ .Values.minio_pull_policy }}
+          env:
+            - name: MC_HOST_local
+              valueFrom:
+                secretKeyRef:
+                  name: storage-auth
+                  key: MC_HOST
+
+          command: ['/bin/sh']
+          args: ['-c', 'mc mb local/test-bucket; mc policy set public local/test-bucket' ]
+{{- end }}
+
       containers:
         - name: api
           image: {{ .Values.api_image }}
@@ -67,10 +86,12 @@ spec:
 
           resources:
             limits:
-              cpu: {{ .Values.api_limit_cpu }}
+              cpu: {{ .Values.api_limits_cpu }}
+              memory: {{ .Values.api_limits_memory }}
 
             requests:
               cpu: {{ .Values.api_requests_cpu }}
+              memory: {{ .Values.api_requests_memory }}
 
 
 ---

chart/values.yaml

@@ -9,9 +9,11 @@ api_password_secret: "c9085f33ecce4347aa1d69339e16c499"
 
 api_num_replicas: 1
 
-api_limit_cpu: "100m"
 api_requests_cpu: "25m"
+api_limits_cpu: "100m"
 
+api_requests_memory: "100M"
+api_limits_memory: "256M"
 
 
 # MongoDB Image
@@ -66,7 +68,11 @@ storage:
 # set to true to use a local minio image
 minio_local: True
 
+minio_scheme: "http"
+minio_host: "local-minio.default:9000"
+
 minio_image: minio/minio
+minio_mc_image: minio/mc
 minio_pull_policy: "IfNotPresent"