storage: use s3v4 signature for presigning urls

backend/btrixcloud/models.py

@@ -1666,6 +1666,7 @@ class S3StorageIn(BaseModel):
     endpoint_url: str
     bucket: str
     access_endpoint_url: Optional[str] = None
+    access_addressing_style: Literal["virtual", "path"] = "virtual"
     region: str = ""
 
 
@@ -1680,6 +1681,7 @@ class S3Storage(BaseModel):
     access_key: str
     secret_key: str
     access_endpoint_url: str
+    access_addressing_style: Literal["virtual", "path"] = "virtual"
     region: str = ""
 
 

backend/btrixcloud/storages.py

@@ -70,7 +70,7 @@
 
 
 # ============================================================================
-# pylint: disable=broad-except,raise-missing-from
+# pylint: disable=broad-except,raise-missing-from,too-many-instance-attributes
 class StorageOps:
     """All storage handling, download/upload operations"""
 
@@ -104,6 +104,8 @@ def __init__(self, org_ops, crawl_manager, mdb) -> None:
         default_namespace = os.environ.get("DEFAULT_NAMESPACE", "default")
         self.frontend_origin = f"{frontend_origin}.{default_namespace}"
 
+        self.local_minio_access_path = os.environ.get("LOCAL_MINIO_ACCESS_PATH")
+
         with open(os.environ["STORAGES_JSON"], encoding="utf-8") as fh:
             storage_list = json.loads(fh.read())
 
@@ -158,13 +160,18 @@ def _create_s3_storage(self, storage: dict[str, str]) -> S3Storage:
 
         access_endpoint_url = storage.get("access_endpoint_url") or endpoint_url
 
+        addressing_style = storage.get("access_addressing_style", "virtual")
+        if access_endpoint_url == self.local_minio_access_path:
+            addressing_style = "path"
+
         return S3Storage(
             access_key=storage["access_key"],
             secret_key=storage["secret_key"],
             region=storage.get("region", ""),
             endpoint_url=endpoint_url,
             endpoint_no_bucket_url=endpoint_no_bucket_url,
             access_endpoint_url=access_endpoint_url,
+            access_addressing_style=addressing_style,
         )
 
     async def add_custom_storage(
@@ -189,6 +196,7 @@ async def add_custom_storage(
             endpoint_url=endpoint_url,
             endpoint_no_bucket_url=endpoint_no_bucket_url,
             access_endpoint_url=storagein.access_endpoint_url or storagein.endpoint_url,
+            access_addressing_style=storagein.access_addressing_style,
         )
 
         try:
@@ -291,9 +299,12 @@ async def get_s3_client(
 
         session = aiobotocore.session.get_session()
 
-        config = None
+        s3 = None
+
         if for_presign and storage.access_endpoint_url != storage.endpoint_url:
-            config = AioConfig(s3={"addressing_style": "virtual"})
+            s3 = {"addressing_style": storage.access_addressing_style}
+
+        config = AioConfig(signature_version="s3v4", s3=s3)
 
         async with session.create_client(
             "s3",
@@ -498,9 +509,12 @@ async def get_presigned_url(
                 s3storage.access_endpoint_url
                 and s3storage.access_endpoint_url != s3storage.endpoint_url
             ):
+                virtual = s3storage.access_addressing_style == "virtual"
                 parts = urlsplit(s3storage.endpoint_url)
                 host_endpoint_url = (
                     f"{parts.scheme}://{bucket}.{parts.netloc}/{orig_key}"
+                    if virtual
+                    else f"{parts.scheme}://{parts.netloc}/{bucket}/{orig_key}"
                 )
                 presigned_url = presigned_url.replace(
                     host_endpoint_url, s3storage.access_endpoint_url

chart/templates/configmap.yaml

@@ -14,7 +14,7 @@ data:
 
   FRONTEND_ORIGIN: {{ .Values.frontend_alias | default "http://browsertrix-cloud-frontend" }}
 
-  CRAWLER_FQDN_SUFFIX: ".{{ .Values.crawler_namespace }}.svc.cluster.local"
+  CRAWLER_FQDN_SUFFIX: ".{{ .Values.crawler_namespace }}{{ .Values.fqdn_suffix }}"
 
   DEFAULT_ORG: "{{ .Values.default_org }}"
 
@@ -53,6 +53,8 @@ data:
 
   IS_LOCAL_MINIO: "{{ .Values.minio_local }}"
 
+  LOCAL_MINIO_ACCESS_PATH: "{{ .Values.minio_access_path }}"
+
   STORAGES_JSON: "/ops-configs/storages.json"
 
   CRAWLER_CHANNELS_JSON: "/ops-configs/crawler_channels.json"

chart/templates/frontend.yaml

@@ -41,7 +41,7 @@ spec:
               value: {{ .Values.name }}-backend
 
             - name: CRAWLER_FQDN_SUFFIX
-              value: ".{{ .Values.crawler_namespace }}.svc.cluster.local"
+              value: ".{{ .Values.crawler_namespace }}{{ .Values.fqdn_suffix }}"
 
             - name: NGINX_ENTRYPOINT_WORKER_PROCESSES_AUTOTUNE
               value: "1"
@@ -60,7 +60,10 @@ spec:
 
             - name: LOCAL_BUCKET
               value: "{{ .Values.minio_local_bucket_name }}"
-            {{- end }}
+
+            - name: LOCAL_ACCESS_PATH
+              value: "{{ .Values.minio_access_path }}"
+             {{- end }}
 
             {{- if .Values.inject_extra }}
             - name: INJECT_EXTRA

chart/templates/minio.yaml

@@ -136,6 +136,23 @@ spec:
       {{- end }}
       name: minio
 
+---
+apiVersion: v1
+kind: Service
+
+metadata:
+  namespace: {{ .Values.crawler_namespace }}
+  name: local-minio
+  labels:
+    app: local-minio
+
+spec:
+  type: ExternalName
+  externalName: "local-minio.{{ .Release.Namespace }}{{ .Values.fqdn_suffix }}"
+  ports:
+    - port: 9000
+
+
 {{- if .Values.minio_local_console_port }}
 ---
 apiVersion: v1

chart/values.yaml

@@ -398,6 +398,9 @@ minio_pull_policy: "IfNotPresent"
 
 minio_local_bucket_name: &local_bucket_name "btrix-data"
 
+# path for serving from local minio bucket
+minio_access_path: &minio_access_path "/data/"
+
 minio_cpu: "10m"
 minio_memory: "1024Mi"
 
@@ -413,8 +416,8 @@ storages:
     secret_key: "PASSW0RD"
     bucket_name: *local_bucket_name
 
-    endpoint_url: "http://local-minio.default:9000/"
-    access_endpoint_url: "/data/"
+    endpoint_url: "http://local-minio:9000/"
+    access_endpoint_url: *minio_access_path
 
 
 # optional: duration in minutes for WACZ download links to be valid
@@ -495,6 +498,9 @@ signer_memory: "50Mi"
 # Other Settings
 # =========================================
 
+# default FQDN suffix, shouldn't need to change
+fqdn_suffix: .svc.cluster.local
+
 # Optional: configure load balancing annotations
 # service:
 #   annotations:

frontend/00-browsertrix-nginx-init.sh

@@ -7,7 +7,9 @@ if [ -z "$LOCAL_MINIO_HOST" ]; then
   echo "no local minio, clearing out minio route"
   echo "" >/etc/nginx/includes/minio.conf
 else
-  echo "local minio: replacing \$LOCAL_MINIO_HOST with \"$LOCAL_MINIO_HOST\", \$LOCAL_BUCKET with \"$LOCAL_BUCKET\""
+  LOCAL_ACCESS_PATH=$(printf '%s\n' "$LOCAL_ACCESS_PATH" | sed -e 's/[\/&]/\\&/g')
+  echo "local minio: replacing \$LOCAL_MINIO_HOST with \"$LOCAL_MINIO_HOST\", \$LOCAL_BUCKET with \"$LOCAL_BUCKET\", \$LOCAL_ACCESS_PATH with \"$LOCAL_ACCESS_PATH\""
+  sed -i "s/\$LOCAL_ACCESS_PATH/$LOCAL_ACCESS_PATH/g" /etc/nginx/includes/minio.conf
   sed -i "s/\$LOCAL_MINIO_HOST/$LOCAL_MINIO_HOST/g" /etc/nginx/includes/minio.conf
   sed -i "s/\$LOCAL_BUCKET/$LOCAL_BUCKET/g" /etc/nginx/includes/minio.conf
 fi

frontend/minio.conf

@@ -1,4 +1,4 @@
-location /data/ {
+location $LOCAL_ACCESS_PATH {
   proxy_pass http://$LOCAL_MINIO_HOST/$LOCAL_BUCKET/;
   proxy_redirect off;
   proxy_buffering off;