Update installer and runtime setup flow

ansible/roles/cezen-backend/files/rag_ingest.py

@@ -20,7 +20,7 @@ Usage (called by main.py):
         --ollama-url http://localhost:11434
 """
 
-import argparse, json, os, sqlite3, sys, uuid
+import argparse, json, logging, os, sqlite3, sys, uuid
 from datetime import datetime, timezone
 from pathlib import Path
 
@@ -38,6 +38,27 @@ CHUNK_SIZE    = 512    # tokens/chars per chunk
 CHUNK_OVERLAP = 64     # overlap between consecutive chunks
 BATCH_SIZE    = 16     # how many chunks to embed + upsert at once
 
+# ── Logging ───────────────────────────────────────────────────────────────────
+# Ingest runs as a detached subprocess of the API, so failures used to be
+# surfaced to the portal as a raw Python traceback stuffed into error_msg.
+# Now the full traceback goes to a log file (server-side only) and the DB gets
+# a short, human-readable message instead.
+_LOG_FILE = Path(args.db_path).resolve().parent / "cezen-ingest.log"
+logger = logging.getLogger("cezen.ingest")
+if not logger.handlers:
+    _handler = logging.FileHandler(_LOG_FILE)
+    _handler.setFormatter(logging.Formatter("%(asctime)s [%(levelname)s] %(message)s"))
+    logger.addHandler(_handler)
+    logger.setLevel(logging.INFO)
+
+def friendly_error(e: Exception, context: str) -> str:
+    """Log the full exception + traceback server-side; return a short, actionable message for the UI."""
+    logger.error("doc_id=%s context=%s", args.doc_id, context, exc_info=True)
+    msg = str(e).strip() or e.__class__.__name__
+    if len(msg) > 200:
+        msg = msg[:200] + "..."
+    return f"{msg} (see server ingest log for full details, request context: {context})"
+
 def utcnow():
     return datetime.now(timezone.utc).isoformat()
 
@@ -256,8 +277,7 @@ def main():
             processed += len(batch)
 
     except Exception as e:
-        import traceback
-        set_status("failed", chunk_count=processed, error=f"{e}\n{traceback.format_exc()[:500]}")
+        set_status("failed", chunk_count=processed, error=friendly_error(e, "embed_and_upsert"))
         sys.exit(1)
 
     set_status("ready", chunk_count=total)
@@ -266,9 +286,8 @@ if __name__ == "__main__":
     try:
         main()
     except KeyboardInterrupt:
-        set_status("failed", error="Interrupted")
+        set_status("failed", error="Ingest was interrupted before it could finish. Try uploading the document again.")
         sys.exit(130)
     except Exception as e:
-        import traceback
-        set_status("failed", error=f"{e}\n{traceback.format_exc()[:500]}")
+        set_status("failed", error=friendly_error(e, "main"))
         sys.exit(1)