Add backup process via honcho Procfile

Dockerfile

@@ -1,4 +1,7 @@
 FROM argilla/argilla-quickstart:latest
 
-# The quickstart image includes Elasticsearch + Argilla
-# Uses default entrypoint - no customization needed for basic setup
+# Copy backup script to a location honcho can find
+COPY backup_to_hub.py /home/argilla/backup_to_hub.py
+
+# Copy custom Procfile that adds backup process
+COPY Procfile /home/argilla/Procfile

Procfile

@@ -0,0 +1,3 @@
+elastic: /usr/share/elasticsearch/bin/elasticsearch
+argilla: sleep 15 && /home/argilla/start_argilla_server.sh
+backup: sleep 60 && python /home/argilla/backup_to_hub.py

backup_to_hub.py

@@ -1,6 +1,6 @@
 """
 Automatic backup of Argilla annotations to HF Dataset.
-Runs as background process, syncs every 10 minutes.
+Runs as a honcho-managed process, syncs every 10 minutes.
 """
 import os
 import sys
@@ -9,8 +9,8 @@ import time
 from pathlib import Path
 from datetime import datetime
 
-# Configuration
-EXPORT_DIR = Path("/exports")
+# Configuration - use writable directory under /home/argilla
+EXPORT_DIR = Path("/home/argilla/exports")
 BACKUP_REPO = os.environ.get("BACKUP_DATASET_REPO", "argus-systems/vuln-preferences-data")
 SYNC_INTERVAL = int(os.environ.get("SYNC_INTERVAL_MINUTES", "10"))
 ARGILLA_API_URL = "http://localhost:6900"
@@ -18,14 +18,14 @@ ARGILLA_API_URL = "http://localhost:6900"
 # Check for HF_TOKEN before importing heavy dependencies
 hf_token = os.environ.get("HF_TOKEN")
 if not hf_token:
-    print("ERROR: HF_TOKEN not set, backup script exiting")
+    print("ERROR: HF_TOKEN not set, backup script exiting", flush=True)
     sys.exit(1)
 
 from huggingface_hub import CommitScheduler, login
 import argilla as rg
 
 # Setup
-EXPORT_DIR.mkdir(exist_ok=True)
+EXPORT_DIR.mkdir(parents=True, exist_ok=True)
 login(token=hf_token)
 
 # Setup CommitScheduler
@@ -38,25 +38,28 @@ scheduler = CommitScheduler(
 
 
 def wait_for_argilla():
-    """Wait for Argilla server to be ready."""
+    """Wait for Argilla server to be ready with extended retries."""
     import urllib.request
-    max_retries = 30
+    max_retries = 60  # Extended retries since we're starting with Argilla
     for i in range(max_retries):
         try:
-            urllib.request.urlopen(f"{ARGILLA_API_URL}/api/v1/status")
-            print("Argilla server is ready")
+            urllib.request.urlopen(f"{ARGILLA_API_URL}/api/v1/status", timeout=5)
+            print("Argilla server is ready", flush=True)
             return True
-        except Exception:
-            print(f"Waiting for Argilla... ({i+1}/{max_retries})")
-            time.sleep(2)
+        except Exception as e:
+            print(f"Waiting for Argilla... ({i+1}/{max_retries}) - {e}", flush=True)
+            time.sleep(5)
+    print("ERROR: Argilla server not responding after maximum retries", flush=True)
     return False
 
 
 def export_annotations():
     """Export all annotations from Argilla to JSON files."""
     try:
-        api_key = os.environ.get("ARGILLA_API_KEY", "argilla.apikey")
+        # Default API key matches what argilla-quickstart creates for the owner user
+        api_key = os.environ.get("ARGILLA_API_KEY", "owner.apikey")
         client = rg.Argilla(api_url=ARGILLA_API_URL, api_key=api_key)
+        print(f"Connected to Argilla, checking datasets...", flush=True)
 
         for dataset in client.datasets:
             records = list(dataset.records)
@@ -74,19 +77,28 @@ def export_annotations():
                             "num_annotations": len(annotated),
                             "records": annotated
                         }, f, indent=2, default=str)
-                print(f"Exported {len(annotated)} annotations from {dataset.name}")
+                print(f"Exported {len(annotated)} annotations from {dataset.name}", flush=True)
+            else:
+                print(f"No new annotations in {dataset.name}", flush=True)
     except Exception as e:
-        print(f"Export error: {e}")
+        print(f"Export error: {e}", flush=True)
+        import traceback
+        traceback.print_exc()
 
 
 def main():
-    print("Starting Argilla backup service...")
+    print("Starting Argilla backup service...", flush=True)
+    print(f"  Backup repo: {BACKUP_REPO}", flush=True)
+    print(f"  Export dir: {EXPORT_DIR}", flush=True)
+    print(f"  Sync interval: {SYNC_INTERVAL} minutes", flush=True)
 
     if not wait_for_argilla():
-        print("Warning: Argilla not responding, will retry exports")
+        print("Warning: Argilla not responding, will retry exports anyway", flush=True)
 
+    print("Starting backup loop...", flush=True)
     while True:
         export_annotations()
+        print(f"Sleeping for {SYNC_INTERVAL} minutes until next export...", flush=True)
         time.sleep(SYNC_INTERVAL * 60)  # Convert to seconds