discours.io/core
5
0
Fork 0
Code Pull Requests Actions Projects Activity

debug: server.py -> threds 1 , search.py -> add 3 times reconect
All checks were successful
Deploy on push / deploy (push) Successful in 49s
Details

Browse Source
This commit is contained in:
Stepan Vladovskiy 2025-03-24 20:16:07 -03:00
parent 60a13a9097
commit 077cb46482
2 changed files with 44 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
server.py
View File

@ -17,7 +17,7 @@ if __name__ == "__main__":
address="0.0.0.0", address="0.0.0.0",
port=PORT, port=PORT,
interface=Interfaces.ASGI, interface=Interfaces.ASGI,
workers=4, workers=1,
websockets=False, websockets=False,
log_level=LogLevels.debug, log_level=LogLevels.debug,
backlog=2048, backlog=2048,

68
services/search.py
View File

@ -290,32 +290,50 @@ class SearchService:
async def _process_single_batch(self, documents, batch_id): async def _process_single_batch(self, documents, batch_id):
"""Process a single batch with maximum reliability""" """Process a single batch with maximum reliability"""
try: max_retries = 3
if not documents: retry_count = 0
return
while retry_count < max_retries:
try:
if not documents:
return
logger.info(f"Processing sub-batch {batch_id} with {len(documents)} documents")
response = await self.index_client.post(
"/bulk-index",
json=documents,
timeout=90.0
)
response.raise_for_status()
result = response.json()
logger.info(f"Sub-batch {batch_id} indexed successfully: {result}")
return # Success, exit the retry loop
logger.info(f"Processing sub-batch {batch_id} with {len(documents)} documents") except Exception as e:
response = await self.index_client.post( error_str = str(e).lower()
"/bulk-index", retry_count += 1
json=documents,
timeout=90.0 # Check if it's a transient error that txtai might recover from internally
) if "dictionary changed size" in error_str or "transaction error" in error_str:
response.raise_for_status() wait_time = (2 ** retry_count) + (random.random() * 0.5)
result = response.json() logger.warning(f"Transient txtai error in sub-batch {batch_id}, waiting {wait_time:.1f}s for recovery: {str(e)[:200]}")
logger.info(f"Sub-batch {batch_id} indexed successfully: {result}") await asyncio.sleep(wait_time) # Wait for txtai to recover
except Exception as e: continue # Try again
logger.error(f"Error indexing sub-batch {batch_id}: {str(e)[:200]}")
# For other errors or final retry failure
# For tiny batches, try one-by-one as last resort logger.error(f"Error indexing sub-batch {batch_id} (attempt {retry_count}/{max_retries}): {str(e)[:200]}")
if len(documents) > 1:
logger.info(f"Processing documents in sub-batch {batch_id} individually") # Only try one-by-one on the final retry
for i, doc in enumerate(documents): if retry_count >= max_retries and len(documents) > 1:
try: logger.info(f"Processing documents in sub-batch {batch_id} individually")
resp = await self.index_client.post("/index", json=doc, timeout=30.0) for i, doc in enumerate(documents):
resp.raise_for_status() try:
logger.info(f"Indexed document {doc['id']} individually") resp = await self.index_client.post("/index", json=doc, timeout=30.0)
except Exception as e2: resp.raise_for_status()
logger.error(f"Failed to index document {doc['id']} individually: {str(e2)[:100]}") logger.info(f"Indexed document {doc['id']} individually")
except Exception as e2:
logger.error(f"Failed to index document {doc['id']} individually: {str(e2)[:100]}")
return # Exit after individual processing attempt
def _truncate_error_detail(self, error_detail): def _truncate_error_detail(self, error_detail):
"""Truncate error details for logging""" """Truncate error details for logging"""