Misc backend fixes for cloud deployment (#26)

* misc backend fixes:
- fix running w/o local minio
- ensure crawler image pull policy is configurable, loaded via chart value
- use digitalocean repo for main backend image (for now)
- add bucket_name to config only if using default bucket

* enable all behaviors, support 'access_endpoint_url' for default storages

* debugging: add 'no_delete_jobs' setting for k8s and docker to disable deletion of completed jobs

backend/crawlconfigs.py

@@ -66,7 +66,7 @@ class RawCrawlConfig(BaseModel):
     combineWARC: Optional[bool] = False
 
     logging: Optional[str] = ""
-    behaviors: Optional[str] = "autoscroll"
+    behaviors: Optional[str] = "autoscroll,autoplay,autofetch,siteSpecific"
 
 
 # ============================================================================

backend/dockerman.py

@@ -44,6 +44,8 @@ class DockerManager:
         self.extra_crawl_params = extra_crawl_params or []
         self._event_q = None
 
+        self.no_delete_on_fail = os.environ.get("NO_DELETE_ON_FAIL", "")
+
         self.storages = {
             "default": S3Storage(
                 name="default",
@@ -111,7 +113,8 @@ class DockerManager:
 
             for container in results:
                 print(f"Cleaning Up Orphan Container {container['Id']}", flush=True)
-                await container.delete()
+                if not self.no_delete_on_fail:
+                    await container.delete()
 
             results = await self.client.containers.list(
                 filters=json.dumps(
@@ -482,8 +485,10 @@ class DockerManager:
         if actor["Attributes"]["exitCode"] != 0:
             crawl = self._make_crawl_for_container(container, "failed", True)
             await self.crawl_ops.store_crawl(crawl)
-
-        await container.delete()
+            if not self.no_delete_on_fail:
+                await container.delete()
+        else:
+            await container.delete()
 
     # pylint: disable=no-self-use,too-many-arguments
     def _make_crawl_for_container(self, container, state, finish_now=False):

backend/k8sman.py

@@ -37,11 +37,13 @@ class K8SManager:
         self.namespace = namespace
         self._default_storage_endpoints = {}
 
-        self.crawler_image = os.environ.get("CRAWLER_IMAGE")
-        self.crawler_image_pull_policy = "IfNotPresent"
+        self.crawler_image = os.environ["CRAWLER_IMAGE"]
+        self.crawler_image_pull_policy = os.environ["CRAWLER_PULL_POLICY"]
 
         self.crawl_retries = int(os.environ.get("CRAWL_RETRIES", "3"))
 
+        self.no_delete_jobs = os.environ.get("NO_DELETE_JOBS", "0") != "0"
+
         self.loop = asyncio.get_running_loop()
         self.loop.create_task(self.run_event_loop())
 
@@ -319,7 +321,7 @@ class K8SManager:
             return None, None
 
         manual = job.metadata.annotations.get("btrix.run.manual") == "1"
-        if manual:
+        if manual and not self.no_delete_jobs:
             self.loop.create_task(self._delete_job(job.metadata.name))
 
         crawl = self._make_crawl_for_job(
@@ -457,7 +459,7 @@ class K8SManager:
         failure = await self.crawl_ops.store_crawl(crawl)
 
         # keep failed jobs around, for now
-        if not failure:
+        if not failure and not self.no_delete_jobs:
             await self._delete_job(job_name)
 
     # ========================================================================
@@ -643,7 +645,7 @@ class K8SManager:
                             {
                                 "name": "crawler",
                                 "image": self.crawler_image,
-                                "imagePullPolicy": "Never",
+                                "imagePullPolicy": self.crawler_image_pull_policy,
                                 "command": [
                                     "crawl",
                                     "--config",

chart/templates/configmap.yaml

@@ -10,6 +10,7 @@ data:
 
   CRAWLER_NAMESPACE: {{ .Values.crawler_namespace }}
   CRAWLER_IMAGE: {{ .Values.crawler_image }}
+  CRAWLER_PULL_POLICY: {{ .Values.crawler_pull_policy }}
 
   CRAWL_TIMEOUT: "{{ .Values.crawl_timeout }}"
   CRAWL_RETRIES: "{{ .Values.crawl_retries }}"
@@ -18,6 +19,8 @@ data:
 
   REDIS_CRAWLS_DONE_KEY: "crawls-done"
 
+  NO_DELETE_JOBS: "{{ .Values.no_delete_jobs | default '0' }}"
+
 ---
 apiVersion: v1
 kind: ConfigMap

chart/templates/ingress.yaml

@@ -29,8 +29,8 @@ spec:
   rules:
   - host: {{ .Values.ingress.host }}
     http:
-{{- if .Values.minio_local }}
       paths:
+{{- if .Values.minio_local }}
       - path: /data/(.*)
         pathType: Prefix
         backend:
@@ -51,7 +51,7 @@ spec:
 {{ if .Values.ingress.tls }}
 ---
 
-apiVersion: cert-manager.io/v1alpha2
+apiVersion: cert-manager.io/v1
 kind: ClusterIssuer
 metadata:
   name: cert-main

chart/templates/main.yaml

@@ -31,8 +31,8 @@ spec:
         - name: nginx-resolver
           emptyDir: {}
 
-{{- if .Values.minio_local }}
       initContainers:
+{{- if .Values.minio_local }}
         - name: init-bucket
           image: {{ .Values.minio_mc_image }}
           imagePullPolicy: {{ .Values.minio_pull_policy }}
@@ -44,7 +44,7 @@ spec:
                   key: MC_HOST
 
           command: ['/bin/sh']
-          args: ['-c', 'mc mb local/test-bucket; mc policy set public local/test-bucket' ]
+          args: ['-c', 'mc mb --ignore-existing local/{{ .Values.minio_local_bucket_name }}' ]
 {{- end }}
 
         - name: init-nginx

chart/templates/secrets.yaml

@@ -31,9 +31,19 @@ type: Opaque
 stringData:
   STORE_ACCESS_KEY: "{{ $storage.access_key }}"
   STORE_SECRET_KEY: "{{ $storage.secret_key }}"
-  STORE_ENDPOINT_URL: "{{ $storage.endpoint_url }}{{ $storage.bucket_name }}/"
-  {{- if and $.Values.ingress.host $.Values.minio_local }}
+
+  {{- if $storage.bucket_name }}
+  STORE_ENDPOINT_URL: "{{ $storage.endpoint_url }}{{ $storage.bucket_name }}"
+  {{- else }}
+  STORE_ENDPOINT_URL: "{{ $storage.endpoint_url }}"
+  {{- end }}
+
+  {{- if $storage.access_endpoint_url }}
+  STORE_ACCESS_ENDPOINT_URL: "{{ $storage.access_endpoint_url }}"
+  {{- else if and $.Values.ingress.host $.Values.minio_local }}
   STORE_ACCESS_ENDPOINT_URL: {{ $.Values.ingress.scheme | default "https" }}://{{ $.Values.ingress.host }}/data/{{ $storage.bucket_name }}/
+  {{- else }}
+  STORE_ACCESS_ENDPOINT_URL: "{{ $storage.endpoint_url }}"
   {{- end }}
 
 {{- end }}

chart/values.yaml

@@ -57,7 +57,7 @@ redis_url: "redis://local-redis.default:6379/1"
 # =========================================
 
 crawler_image: "webrecorder/browsertrix-crawler:latest"
-crawler_pull_policy: "Never"
+crawler_pull_policy: "IfNotPresent"
 
 crawler_namespace: "crawlers"
 
@@ -68,19 +68,6 @@ crawl_retries: 1
 crawler_args: "--timeout 90 --logging stats,behaviors,debug --generateWACZ --screencastPort 9037"
 
 
-
-# Storage
-# =========================================
-
-storages:
-  - name: "default"
-    access_key: "ADMIN"
-    secret_key: "PASSW0RD"
-    bucket_name: "test-bucket"
-
-    endpoint_url: "http://local-minio.default:9000/"
-
-
 # Local Minio Pod (optional)
 # =========================================
 # set to true to use a local minio image
@@ -93,6 +80,21 @@ minio_image: minio/minio
 minio_mc_image: minio/mc
 minio_pull_policy: "IfNotPresent"
 
+minio_local_bucket_name: &local_bucket_name "test-bucket"
+
+
+# Storage
+# =========================================
+# should include the local minio bucket, if enabled, and any other available buckets for default storage
+
+storages:
+  - name: "default"
+    access_key: "ADMIN"
+    secret_key: "PASSW0RD"
+    bucket_name: *local_bucket_name
+
+    endpoint_url: "http://local-minio.default:9000/"
+
 
 # Deployment options
 # =========================================

docker-compose.yml

@@ -3,7 +3,7 @@ version: '3.5'
 services:
   backend:
     build: ./backend
-    image: webrecorder/browsertrix-api
+    image: registry.digitalocean.com/btrix/webrecorder/browsertrix-api
     ports:
       - 8000:8000