PromtEngineer
Add Document from localGPT UI is not working
When I click on Upload and click on Add button it is throwing: DB\chroma.sqlite3 - The process cannot access the file because it is being used by another process.
I'm getting the same error on Windows trying to ADD or RESET from the Web UI
INFO:werkzeug:WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
* Running on http://127.0.0.1:5110
INFO:werkzeug:Press CTRL+C to quit
ERROR:run_localGPT_API:Exception on /api/delete_source [GET]
Traceback (most recent call last):
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 1455, in wsgi_app
response = self.full_dispatch_request()
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 869, in full_dispatch_request
rv = self.handle_user_exception(e)
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 867, in full_dispatch_request
rv = self.dispatch_request()
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 852, in dispatch_request
return self.ensure_sync(self.view_functions[rule.endpoint])(**view_args)
File "D:\AI\localgpt\run_localGPT_API.py", line 85, in delete_source_route
shutil.rmtree(folder_name)
File "C:\Users\talle\anaconda3\envs\localgpt\lib\shutil.py", line 750, in rmtree
return _rmtree_unsafe(path, onerror)
File "C:\Users\talle\anaconda3\envs\localgpt\lib\shutil.py", line 624, in _rmtree_unsafe
onerror(os.rmdir, path, sys.exc_info())
File "C:\Users\talle\anaconda3\envs\localgpt\lib\shutil.py", line 622, in _rmtree_unsafe
os.rmdir(path)
PermissionError: [WinError 32] The process cannot access the file because it is being used by another process: 'SOURCE_DOCUMENTS'
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:44:51] "GET /api/delete_source HTTP/1.1" 500 -
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:44:53] "POST /api/save_document HTTP/1.1" 200 -
Error: D:\AI\localgpt/DB\chroma.sqlite3 - The process cannot access the file because it is being used by another process.
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:45:19] "GET /api/run_ingest HTTP/1.1" 500 -
ERROR:run_localGPT_API:Exception on /api/delete_source [GET]
Traceback (most recent call last):
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 1455, in wsgi_app
response = self.full_dispatch_request()
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 869, in full_dispatch_request
rv = self.handle_user_exception(e)
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 867, in full_dispatch_request
rv = self.dispatch_request()
File "C:\Users\talle\anaconda3\envs\localgpt\lib\site-packages\flask\app.py", line 852, in dispatch_request
return self.ensure_sync(self.view_functions[rule.endpoint])(**view_args)
File "D:\AI\localgpt\run_localGPT_API.py", line 85, in delete_source_route
shutil.rmtree(folder_name)
File "C:\Users\talle\anaconda3\envs\localgpt\lib\shutil.py", line 750, in rmtree
return _rmtree_unsafe(path, onerror)
File "C:\Users\talle\anaconda3\envs\localgpt\lib\shutil.py", line 624, in _rmtree_unsafe
onerror(os.rmdir, path, sys.exc_info())
File "C:\Users\talle\anaconda3\envs\localgpt\lib\shutil.py", line 622, in _rmtree_unsafe
os.rmdir(path)
PermissionError: [WinError 32] The process cannot access the file because it is being used by another process: 'SOURCE_DOCUMENTS'
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:46:57] "GET /api/delete_source HTTP/1.1" 500 -
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:46:59] "POST /api/save_document HTTP/1.1" 200 -
Error: D:\AI\localgpt/DB\chroma.sqlite3 - The process cannot access the file because it is being used by another process.
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:47:23] "GET /api/run_ingest HTTP/1.1" 200 -
llama_print_timings: load time = 468.10 ms
llama_print_timings: sample time = 31.59 ms / 217 runs ( 0.15 ms per token, 6869.48 tokens per second)
llama_print_timings: prompt eval time = 1253.37 ms / 1267 tokens ( 0.99 ms per token, 1010.87 tokens per second)
llama_print_timings: eval time = 6252.20 ms / 216 runs ( 28.95 ms per token, 34.55 tokens per second)
llama_print_timings: total time = 7895.91 ms
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:48:00] "POST /api/prompt_route HTTP/1.1" 200 -
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:50:41] "POST /api/save_document HTTP/1.1" 200 -
Error: D:\AI\localgpt/DB\79613b6b-03e9-494a-805a-15528e0c2992\data_level0.bin - The process cannot access the file because it is being used by another process.
INFO:werkzeug:127.0.0.1 - - [17/Oct/2023 15:51:07] "GET /api/run_ingest HTTP/1.1" 200 -
Same problem on my end. I'm new to this project, so I don't really know where to start with debugging it.
Same here. while the "run_localGPT_API.py" is running, the UI part can not access to the DB.
Hello everybody, I have found a possible solution (although not a very elegant one). The problem is that the database file should be deleted by shutil.rmtree() while still 'in use'. To make sure the file is not used anymore and can be deleted, simply write
DB = None
RETRIEVER = None
QA = None
gc.collect()
before the execution of shutil.rmtree(). This sets all variables to None and then calls the python garbage collector explicitly (don't know if this is needed in any case, but better safe than sorry). Additionnally, I replaced the line result = subprocess.run(run_langest_commands, capture_output=True) by explicit calls of the ingest functions. In this way, it is assured that the database is only accessed on the main thread.
Hello everybody, I have found a possible solution (although not a very elegant one). The problem is that the database file should be deleted by shutil.rmtree() while still 'in use'. To make sure the file is not used anymore and can be deleted, simply write
DB = None RETRIEVER = None QA = None gc.collect()before the execution of shutil.rmtree(). This sets all variables to None and then calls the python garbage collector explicitly (don't know if this is needed in any case, but better safe than sorry). Additionnally, I replaced the line
result = subprocess.run(run_langest_commands, capture_output=True)by explicit calls of the ingest functions. In this way, it is assured that the database is only accessed on the main thread.
Hi cassimon, Do you mind share the modified code? I am having a hard time trying to figure it out. Thank You
No problem. But it will expose that the solution is really not ideal, which is why I am not very proud of it. I modified the ingest response in the following way
global DB
global RETRIEVER
global QA
try:
if os.path.exists(PERSIST_DIRECTORY):
numberOfAttempts = 0;
maxNumberOfAttempts = 10;
while(True):
try:
DB = None;
RETRIEVER = None;
QA = None;
gc.collect();
shutil.rmtree(PERSIST_DIRECTORY)
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(filename)s:%(lineno)s - %(message)s", level=logging.INFO
)
ingest.test("cpu") #Ingest in main thread
print("Successfully loaded files")
break;
except Exception as e:
print("Attempt Failed, retry in 1 sec")
time.sleep(1);
numberOfAttempts += 1;
if(numberOfAttempts==maxNumberOfAttempts):
print("Ingestion failed after max number of attempts ",e)
return "Error", 500;
#run_langest_commands = ["python", "ingest.py"]
#if DEVICE_TYPE == "cpu":
# run_langest_commands.append("--device_type")
# run_langest_commands.append(DEVICE_TYPE)
#f result.returncode != 0:
#return "Script execution failed: {}".format(result.stderr.decode("utf-8")), 500
# load the vectorstore
DB = Chroma(
persist_directory=PERSIST_DIRECTORY,
embedding_function=EMBEDDINGS,
client_settings=CHROMA_SETTINGS,
)
RETRIEVER = DB.as_retriever()
prompt, memory = get_prompt_template(promptTemplate_type="llama", history=False)
QA = RetrievalQA.from_chain_type(
llm=LLM,
chain_type="stuff",
retriever=RETRIEVER,
return_source_documents=SHOW_SOURCES,
chain_type_kwargs={
"prompt": prompt,
},
)
return "Script executed successfully: ", 200
except Exception as e:
print(f"Error occurred: {str(e)}")
return f"Error occurred: {str(e)}", 500
The reason for executing multiple attempts is that sometimes, the parallel execution of ProcessPoolExecutor() fails. I have not figured out why this happens and if this is related to the earlier problem or not. The test() function in ingest.py is implemented as follows.
def test(device_type):
# Load documents and split in chunks
logging.info(f"Loading documents from {SOURCE_DIRECTORY}")
documents = load_documents(SOURCE_DIRECTORY)
text_documents, python_documents = split_documents(documents)
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
python_splitter = RecursiveCharacterTextSplitter.from_language(
language=Language.PYTHON, chunk_size=880, chunk_overlap=200
)
texts = text_splitter.split_documents(text_documents)
texts.extend(python_splitter.split_documents(python_documents))
logging.info(f"Loaded {len(documents)} documents from {SOURCE_DIRECTORY}")
logging.info(f"Split into {len(texts)} chunks of text")
# Create embeddings
embeddings = HuggingFaceInstructEmbeddings(
model_name=EMBEDDING_MODEL_NAME,
model_kwargs={"device": device_type},
)
db = Chroma.from_documents(
texts,
embeddings,
persist_directory=PERSIST_DIRECTORY,
client_settings=CHROMA_SETTINGS,
)
Thank you, I'll test that out and let you know
Sent with Tuta; enjoy secure & ad-free emails: https://tuta.com
17 Jan 2024, 14:08 by @.***:
No problem. But it will expose that the solution is really not ideal, whichis why I am not very proud of it. I modified the ingest response in the following way
if os.path.exists(PERSIST_DIRECTORY): numberOfAttempts = 0; maxNumberOfAttempts = 10; while(True): try: DB = None; RETRIEVER = None; QA = None; gc.collect(); shutil.rmtree(PERSIST_DIRECTORY) logging.basicConfig( format="%(asctime)s - %(levelname)s - %(filename)s:%(lineno)s - %(message)s", level=logging.INFO ) ingest.test("cpu") #Ingest in main thread print("Successfully loaded files") break; except Exception as e: print("Attempt Failed, retry in 1 sec") time.sleep(1); numberOfAttempts += 1; if(numberOfAttempts==maxNumberOfAttempts): print("Ingestion failed after max number of attempts ",e) return "Error", 500;The reason for executing multiple attempts is that sometimes, the parallel execution of ProcessPoolExecutor() fails. I have not figured out why this happens and if this is related to the earlier problem or not. The test() function in ingest.py is implemented as follows.
def test(device_type): # Load documents and split in chunks logging.info(f"Loading documents from {SOURCE_DIRECTORY}") documents = load_documents(SOURCE_DIRECTORY) text_documents, python_documents = split_documents(documents) text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200) python_splitter = RecursiveCharacterTextSplitter.from_language( language=Language.PYTHON, chunk_size=880, chunk_overlap=200 ) texts = text_splitter.split_documents(text_documents) texts.extend(python_splitter.split_documents(python_documents)) logging.info(f"Loaded {len(documents)} documents from {SOURCE_DIRECTORY}") logging.info(f"Split into {len(texts)} chunks of text") # Create embeddings embeddings = HuggingFaceInstructEmbeddings( model_name=EMBEDDING_MODEL_NAME, model_kwargs={"device": device_type}, ) db = Chroma.from_documents( texts, embeddings, persist_directory=PERSIST_DIRECTORY, client_settings=CHROMA_SETTINGS, )
— Reply to this email directly, > view it on GitHub https://github.com/PromtEngineer/localGPT/issues/587#issuecomment-1895777596> , or > unsubscribe https://github.com/notifications/unsubscribe-auth/BCX4VW4X3YAP637YAGAK3JDYO7EMPAVCNFSM6AAAAAA6BX6UZSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOJVG43TONJZGY> . You are receiving this because you are subscribed to this thread.> Message ID: > <PromtEngineer/localGPT/issues/587/1895777596> @> github> .> com>
