add S3 policy docs to README

.env.example

@@ -1,5 +1,6 @@
 S3_ENDPOINT=seaweedfs:8333
 S3_PUBLIC_ENDPOINT=localhost:8333
+S3_BUCKET=officeconvert
 S3_USE_SSL=false
 # Presigned URLs; omit to match S3_USE_SSL (internal client uses S3_ENDPOINT).
 S3_PUBLIC_USE_SSL=false

Makefile

@@ -41,6 +41,7 @@ run-server:
 	if [ "$${S3_PUBLIC_ENDPOINT:-}" = "seaweedfs:8333" ]; then S3_PUBLIC_ENDPOINT=localhost:8333; fi; \
 	export S3_ENDPOINT="$${S3_ENDPOINT:-localhost:8333}"; \
 	export S3_PUBLIC_ENDPOINT="$${S3_PUBLIC_ENDPOINT:-localhost:8333}"; \
+	export S3_BUCKET="$${S3_BUCKET:-officeconvert}"; \
 	export S3_USE_SSL="$${S3_USE_SSL:-false}"; \
 	export S3_ACCESS_KEY="$${S3_ACCESS_KEY:-minioadmin}"; \
 	export S3_SECRET_KEY="$${S3_SECRET_KEY:-minioadmin}"; \

README.md

@@ -135,9 +135,77 @@ Use `.env.example` as your baseline env configuration.
 
 ## Storage Backend Notes
 
-- This project defaults to **SeaweedFS S3 API** for object transit in development and compose deployments.
+- Local development defaults to **SeaweedFS** (S3-compatible) via Docker Compose.
-- The Python server uses the `minio` Python SDK, which is intentional because SeaweedFS is S3-compatible.
+- Production can use any S3-compatible provider; **AWS S3** is the expected choice.
+- The Python server uses the `minio` Python SDK against the S3 API.
 - Runtime configuration uses `S3_*` environment variables.
+- All conversions share one bucket (`S3_BUCKET`, required). Each conversion's objects live under a `{conversion_id}/` key prefix (for example `{conversion_id}/input/source.pptx` and `{conversion_id}/output/slide-0001.jpg`).
+
+### AWS setup
+
+**Bucket**
+
+1. Create one bucket (for example `officeconvert-prod`) in the region where the server runs.
+2. Leave **Block Public Access** enabled. Presigned URLs work without a public bucket.
+3. Optional: add a lifecycle rule to expire objects after a few days as a safety net if cleanup fails.
+
+**Server environment**
+
+Set at minimum:
+
+```bash
+S3_BUCKET=officeconvert-prod
+S3_ENDPOINT=s3.us-east-1.amazonaws.com
+S3_PUBLIC_ENDPOINT=s3.us-east-1.amazonaws.com
+S3_REGION=us-east-1
+S3_USE_SSL=true
+S3_PUBLIC_USE_SSL=true
+S3_ACCESS_KEY=...
+S3_SECRET_KEY=...
+```
+
+Use your bucket's regional hostname for both endpoints unless you deliberately split internal vs client-facing access. `S3_PUBLIC_ENDPOINT` must be reachable by whatever uploads and downloads via presigned URLs (clients, not just the server).
+
+On startup the server calls `CreateBucket` if the bucket is missing. In AWS it is simpler to **pre-create the bucket** and grant object permissions only (see IAM below).
+
+**IAM permissions**
+
+Scope access to the single bucket. Object keys are per-conversion prefixes, so list/delete can target the whole bucket:
+
+```json
+{
+  "Version": "2012-10-17",
+  "Statement": [
+    {
+      "Effect": "Allow",
+      "Action": ["s3:ListBucket"],
+      "Resource": "arn:aws:s3:::officeconvert-prod"
+    },
+    {
+      "Effect": "Allow",
+      "Action": ["s3:GetObject", "s3:PutObject", "s3:DeleteObject"],
+      "Resource": "arn:aws:s3:::officeconvert-prod/*"
+    }
+  ]
+}
+```
+
+Add `s3:CreateBucket` on `arn:aws:s3:::officeconvert-prod` only if you want the server to create the bucket on first boot.
+
+**CORS**
+
+Required only if uploads or downloads go **directly from a browser** to presigned URLs. Server-side clients (`curl`, the Go client) do not need CORS. Allow `PUT` and `GET` for your web origin on the bucket.
+
+**IAM roles vs IAM users**
+
+AWS recommends **roles** over long-lived **IAM user** access keys when the server runs on AWS compute (ECS, EC2, Lambda): a role grants **temporary** credentials that rotate automatically, with no static keys to store or leak.
+
+For this project today, the server reads explicit `S3_ACCESS_KEY` and `S3_SECRET_KEY` via the MinIO SDK. That maps cleanly to:
+
+| Where you run | Practical choice |
+|---------------|------------------|
+| Docker on a VPS, bare metal, or outside AWS | IAM **user** with the policy above; store keys in env or a secrets manager. Fine for a single service at low volume. |
+| ECS / EC2 / EKS on AWS | Prefer an IAM **role** attached to the task or instance. Your orchestrator injects short-lived credentials; you still pass them into `S3_ACCESS_KEY` / `S3_SECRET_KEY` (and a session token if your runtime provides one — the server does not yet read a dedicated `S3_SESSION_TOKEN` env var). |
 
 ## Conversion Tuning Notes
 

deploy/docker-compose.yml

@@ -23,6 +23,7 @@ services:
     environment:
       S3_ENDPOINT: ${S3_ENDPOINT:-seaweedfs:8333}
       S3_PUBLIC_ENDPOINT: ${S3_PUBLIC_ENDPOINT:-localhost:8333}
+      S3_BUCKET: ${S3_BUCKET:-officeconvert}
       S3_USE_SSL: ${S3_USE_SSL:-false}
       S3_ACCESS_KEY: ${S3_ACCESS_KEY:-minioadmin}
       S3_SECRET_KEY: ${S3_SECRET_KEY:-minioadmin}

docker-compose.yml

@@ -22,6 +22,7 @@ services:
     environment:
       S3_ENDPOINT: ${S3_ENDPOINT:-seaweedfs:8333}
       S3_PUBLIC_ENDPOINT: ${S3_PUBLIC_ENDPOINT:-localhost:8333}
+      S3_BUCKET: ${S3_BUCKET:-officeconvert}
       S3_USE_SSL: ${S3_USE_SSL:-false}
       S3_ACCESS_KEY: ${S3_ACCESS_KEY:-minioadmin}
       S3_SECRET_KEY: ${S3_SECRET_KEY:-minioadmin}

python/packages/server/src/officeconvert_server/app.py

@@ -10,7 +10,9 @@ from officeconvertapi.v1.conversion_connect import ConversionServiceASGIApplicat
 
 from officeconvert_server.config import load_server_config
 from officeconvert_server.service import ConversionServiceImpl
-from officeconvert_server.storage import S3Store
+from officeconvert_server.storage import S3Store, log_s3_error
+
+from minio.error import S3Error
 
 logger = logging.getLogger(__name__)
 
@@ -55,6 +57,16 @@ def create_app() -> ConversionServiceASGIApplication:
     if os.getenv("OFFICECONVERT_S3_TRACE", "").lower() in ("1", "true", "yes"):
         store.enable_http_trace(sys.stderr)
         logger.warning("OFFICECONVERT_S3_TRACE enabled: S3 HTTP dumps on stderr")
+    try:
+        store.ensure_bucket(config.s3_bucket)
+    except S3Error as exc:
+        log_s3_error(
+            "ensure_bucket",
+            endpoint=config.s3_endpoint,
+            secure=config.s3_secure,
+            exc=exc,
+        )
+        raise
     service = ConversionServiceImpl(config=config, store=store)
     return ConversionServiceASGIApplication(service)
 

python/packages/server/src/officeconvert_server/config.py

@@ -10,6 +10,7 @@ import os
 class ServerConfig:
     """Defines environment-driven settings for server orchestration."""
 
+    s3_bucket: str
     s3_endpoint: str
     s3_access_key: str
     s3_secret_key: str
@@ -37,7 +38,11 @@ def load_server_config() -> ServerConfig:
         else s3_secure
     )
     region_env = os.getenv("S3_REGION", "").strip()
+    s3_bucket = os.getenv("S3_BUCKET", "").strip()
+    if not s3_bucket:
+        raise ValueError("S3_BUCKET is required")
     return ServerConfig(
+        s3_bucket=s3_bucket,
         s3_endpoint=os.getenv("S3_ENDPOINT", "localhost:8333"),
         s3_access_key=os.getenv("S3_ACCESS_KEY", "minioadmin"),
         s3_secret_key=os.getenv("S3_SECRET_KEY", "minioadmin"),

python/packages/server/src/officeconvert_server/models.py

@@ -23,7 +23,7 @@ class ConversionSession:
     thumbnail_resolution: conversion_pb2.ConversionResolution
     full_jpeg_quality: int
     thumbnail_jpeg_quality: int
-    bucket_name: str
+    object_prefix: str
     upload_object_key: str
     status: conversion_pb2.ConversionStatus
     notes: conversion_pb2.NotesOptions | None = None

python/packages/server/src/officeconvert_server/service.py

@@ -122,23 +122,13 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
 
         ksuid = Ksuid()
         conversion_id = str(ksuid)
-        bucket_name = f"oc-{bytes(ksuid).hex()}"
+        object_prefix = f"{conversion_id}/"
-        upload_key = "input/source.pptx"
+        upload_key = f"{object_prefix}input/source.pptx"
         expires_at = utc_now() + timedelta(seconds=self._config.s3_session_ttl_seconds)
 
-        try:
-            self._store.ensure_bucket(bucket_name)
-        except S3Error as exc:
-            log_s3_error(
-                "ensure_bucket",
-                endpoint=self._config.s3_endpoint,
-                secure=self._config.s3_secure,
-                exc=exc,
-            )
-            raise
         try:
             upload_url = self._store.presigned_put_url(
-                bucket_name,
+                self._config.s3_bucket,
                 upload_key,
                 ttl_seconds=self._config.s3_session_ttl_seconds,
             )
@@ -159,7 +149,7 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
             full_jpeg_quality=full_jpeg_quality,
             thumbnail_jpeg_quality=thumbnail_jpeg_quality,
             notes=request.notes if request.HasField("notes") else None,
-            bucket_name=bucket_name,
+            object_prefix=object_prefix,
             upload_object_key=upload_key,
             status=conversion_pb2.CONVERSION_STATUS_PENDING,
         )
@@ -168,7 +158,7 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
 
         return conversion_pb2.CreateConversionResponse(
             conversion_id=conversion_id,
-            upload_bucket=bucket_name,
+            upload_bucket=self._config.s3_bucket,
             upload_object_key=upload_key,
             upload_url=upload_url,
             expires_at=_to_timestamp(expires_at),
@@ -265,7 +255,11 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
         if session.conversion_task is not None and not session.conversion_task.done():
             session.conversion_task.cancel()
         await self._cleanup_local_artifacts(session)
-        await asyncio.to_thread(self._store.remove_bucket_tree, session.bucket_name)
+        await asyncio.to_thread(
+            self._store.remove_prefix,
+            self._config.s3_bucket,
+            session.object_prefix,
+        )
         return conversion_pb2.DeleteConversionResponse(
             conversion_id=session.conversion_id,
             deleted=True,
@@ -295,7 +289,7 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
         try:
             await asyncio.to_thread(
                 self._store.fget_object,
-                session.bucket_name,
+                self._config.s3_bucket,
                 session.upload_object_key,
                 source_path,
             )
@@ -436,10 +430,12 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
         upload_total = slide_total * 2
         upload_index = 0
         for slide in slides:
-            object_key = f"output/slide-{slide.index:04d}{slide.image_path.suffix}"
+            object_key = (
-            self._store.fput_object(session.bucket_name, object_key, slide.image_path)
+                f"{session.object_prefix}output/slide-{slide.index:04d}{slide.image_path.suffix}"
+            )
+            self._store.fput_object(self._config.s3_bucket, object_key, slide.image_path)
             image_url = self._store.presigned_get_url(
-                session.bucket_name,
+                self._config.s3_bucket,
                 object_key,
                 ttl_seconds=self._config.s3_session_ttl_seconds,
             )
@@ -447,15 +443,16 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
             if progress_callback is not None:
                 progress_callback(upload_index, upload_total)
             thumbnail_object_key = (
-                f"output/thumb/slide-{slide.index:04d}{slide.thumbnail_path.suffix}"
+                f"{session.object_prefix}output/thumb/slide-{slide.index:04d}"
+                f"{slide.thumbnail_path.suffix}"
             )
             self._store.fput_object(
-                session.bucket_name,
+                self._config.s3_bucket,
                 thumbnail_object_key,
                 slide.thumbnail_path,
             )
             thumbnail_image_url = self._store.presigned_get_url(
-                session.bucket_name,
+                self._config.s3_bucket,
                 thumbnail_object_key,
                 ttl_seconds=self._config.s3_session_ttl_seconds,
             )
@@ -515,7 +512,11 @@ class ConversionServiceImpl(conversion_connect.ConversionService):
         """Delete storage resources after the configured session retention period."""
         try:
             await asyncio.sleep(self._config.conversion_cleanup_delay_seconds)
-            await asyncio.to_thread(self._store.remove_bucket_tree, session.bucket_name)
+            await asyncio.to_thread(
+                self._store.remove_prefix,
+                self._config.s3_bucket,
+                session.object_prefix,
+            )
         except asyncio.CancelledError:
             return
         finally:

python/packages/server/src/officeconvert_server/storage.py

@@ -123,35 +123,38 @@ class S3Store:
         """Upload one local filesystem object to storage."""
         self._client.fput_object(bucket_name, object_key, str(source_path))
 
-    def remove_bucket_tree(self, bucket_name: str) -> None:
+    def remove_prefix(self, bucket_name: str, prefix: str) -> None:
-        """Remove all objects in a bucket and then delete the bucket."""
+        """Remove all objects under a key prefix within a bucket."""
-        objects = list(self._client.list_objects(bucket_name, recursive=True))
+        normalized_prefix = prefix if prefix.endswith("/") else f"{prefix}/"
-        if objects:
+        objects = list(
-            delete_requests: list[DeleteObject] = []
+            self._client.list_objects(
-            for obj in objects:
-                object_name = obj.object_name
-                if object_name is None:
-                    raise RuntimeError(
-                        "encountered unnamed object while removing bucket contents"
-                    )
-                delete_requests.append(DeleteObject(object_name))
-
-            errors = self._client.remove_objects(
                 bucket_name,
-                delete_requests,
+                prefix=normalized_prefix,
+                recursive=True,
             )
-            for err in errors:
+        )
-                object_name = err.name or "<unknown>"
+        if not objects:
-                message = err.message or err.code
+            return
+
+        delete_requests: list[DeleteObject] = []
+        for obj in objects:
+            object_name = obj.object_name
+            if object_name is None:
                 raise RuntimeError(
-                    f"failed to delete object {object_name}: {message}"
+                    "encountered unnamed object while removing prefix contents"
                 )
-        try:
+            delete_requests.append(DeleteObject(object_name))
-            self._client.remove_bucket(bucket_name)
+
-        except S3Error as exc:
+        errors = self._client.remove_objects(
-            # Concurrent cleanup paths may race to remove the same bucket.
+            bucket_name,
-            if exc.code != "NoSuchBucket":
+            delete_requests,
-                raise
+        )
+        for err in errors:
+            object_name = err.name or "<unknown>"
+            message = err.message or err.code
+            raise RuntimeError(
+                f"failed to delete object {object_name}: {message}"
+            )
 
 
 def object_key_from_presigned_url(url: str) -> str: