Mark status of task as ERROR when worker fails

asr.py

@@ -2,6 +2,7 @@
 import os
 import time
 import tomli
+from typing import Optional
 
 from base_util import (
     get_asset_info,
@@ -37,7 +38,7 @@ def _get_project_meta():
 version = str(pkg_meta["version"])
 
 
-def run(input_uri: str, output_uri: str, model=None) -> bool:
+def run(input_uri: str, output_uri: str, model=None) -> Optional[str]:
     logger.info(f"Processing {input_uri} (save to --> {output_uri})")
     start_time = time.time()
     prov_steps = []  # track provenance
@@ -46,7 +47,7 @@ def run(input_uri: str, output_uri: str, model=None) -> bool:
     logger.info(result)
     if not result:
         logger.error("Could not obtain input, quitting...")
-        return False
+        return "Input download failed"
 
     prov_steps.append(result.provenance)
 
@@ -58,7 +59,7 @@ def run(input_uri: str, output_uri: str, model=None) -> bool:
     transcode_output = try_transcode(input_path, asset_id, extension)
     if not transcode_output:
         logger.error("The transcode failed to yield a valid file to continue with")
-        return False
+        return "Transcode failed"
     else:
         input_path = transcode_output.transcoded_file_path
         prov_steps.append(transcode_output.provenance)
@@ -137,7 +138,7 @@ def run(input_uri: str, output_uri: str, model=None) -> bool:
     else:
         logger.info("No output_uri specified, so all is done")
 
-    return True
+    return None
 
 
 # if S3 output_uri is supplied transfers data to S3 location

docker-compose.yml

@@ -5,7 +5,6 @@ services:
       - ./model:/model
       - ./data:/data
     container_name: whisper-asr-worker
-    # args: ["--output-uri"]
     env_file:
       - .env.override
     logging:

main.py

@@ -27,7 +27,7 @@ def run_job(input_uri: str, output_uri: str):
         logger.error("Please supply the --input and --output params")
         return False
 
-    return asr.run(input_uri, output_uri)
+    asr.run(input_uri, output_uri)
 
 
 # Start the worker

whisper_api.py

@@ -41,6 +41,7 @@ class Task(BaseModel):
     output_uri: str
     status: Status = Status.CREATED
     id: str | None = None
+    error_msg: str | None = None
 
 
 all_tasks = [
@@ -91,8 +92,9 @@ def try_whisper(task: Task):
     try:
         task.status = Status.PROCESSING
         update_task(task)
-        run(task.input_uri, task.output_uri, model)
-        task.status = Status.DONE
+        error_msg = run(task.input_uri, task.output_uri, model)
+        task.status = Status.ERROR if error_msg else Status.DONE
+        task.error_msg = error_msg
     except Exception:
         logger.exception("Failed to run whisper")
         task.status = Status.ERROR