Currently this is what my deployed web app looks like when I open the link. When I attached a parameter (necessary for how the app functions) I got a 504.0 Gateway Timeout Error. The parameter is used in regard to an SQL Server Database table query. When the app doesn't have the param it opens up and renders a different html page. The container error looks like this: On my GitHub, the app says it is successfully deployed. The app is a Langchain based Flask application. At the start of the application, the web app is meant to connect to my SQL server, using the parameter, and grab data from some tables and turn them into panda data frames in Python. Locally, this process runs perfectly, however it does not work in the deployment URL. The code I used in my python flask app is: app.run(host="0.0.0.0", port="8000") And in my startup.txt file I use: gunicorn --bind=0.0.0.0 --timeout 600 app:app Both these commands are correct my Flask app is called app and is in app.py. I have checked to make sure all environment variables are correct and exist when I check the SSH. Furthermore, I have env WEBSITES_PORT set to 8000. Previously I was having an issue connecting to my SQL Server, but I received help and set up the connection to the server using VNet integration. I no longer get error messages pertaining to the server but the timeout in regard to the container is persistent. I even increased the timeout time to be 600 seconds instead of 230 and it still failed. I am not using dockers in any way in my project either. I am stuck and not sure how to get rid of this error. Again, the web app works extremely well locally so I know this must be a deployment/server issue which is not my forte. Also, In the application logs, there are no erors, the web app is listening on port 0.0.0.0.8000 and reading my startup.txt file. If there's any other information, I can provide for clarity, please let me know! I would really appreciate any and all help! Edit: Using the diagnostic helps in the web app it recommended I take a look at this page: https://azureossd.github.io/2023/02/15/Whats-the-difference-between-PORT-and-WEBSITES_PORT/index.html However, I have set the port to be 8000 and I still have this issue. Should I maybe use PORT instead? I also don't have any dockers files or anything related in this project so not sure if Dockers plays a role in fixing the 504 Gateway Timeout Error. This is from the log stream that is in the picture below. It seems as if the web app is pinging the port that is used for SQL Server: 2024-03-31T23:59:14.282Z ERROR - Container remi2_1_1e46950a for site remi2 did not start within expected time limit. Elapsed time = 601.0096697 sec 2024-03-31T23:59:14.291Z ERROR - Container remi2_1_1e46950a didn't respond to HTTP pings on port: 1433, failing site start. See container logs for debugging. Here is the diagnostic report:

@Ahad Anjum Based on the information you provided, it looks like the container for your site is not starting within the expected time limit and the app is not responding to health checks. One possible reason for this issue could be that the app is trying to connect to the SQL Server using the wrong port. The error message you provided indicates that the container didn't respond to HTTP pings on port 1433. This port is typically used for SQL Server, so it's possible that the app is trying to connect to the SQL Server using this port instead of the correct port. You mentioned that you have set the port to be 8000, but it's still not working. Have you tried using the PORT environment variable instead of WEBSITES_PORT ? According to the link you provided, WEBSITES_PORT is used for internal communication between the Azure App Service and the container, while PORT is used by the container to listen for incoming requests.

Azure Flask Web App Fails During Deployment (504 Gateway Timeout Error)

Accepted answer

brtrach-MSFT 15,251 Reputation points Microsoft Employee

2024-04-02T00:46:33.4433333+00:00

@Ahad Anjum Based on the information you provided, it looks like the container for your site is not starting within the expected time limit and the app is not responding to health checks.

One possible reason for this issue could be that the app is trying to connect to the SQL Server using the wrong port. The error message you provided indicates that the container didn't respond to HTTP pings on port 1433. This port is typically used for SQL Server, so it's possible that the app is trying to connect to the SQL Server using this port instead of the correct port.

You mentioned that you have set the port to be 8000, but it's still not working. Have you tried using the PORT environment variable instead of WEBSITES_PORT? According to the link you provided, WEBSITES_PORT is used for internal communication between the Azure App Service and the container, while PORT is used by the container to listen for incoming requests.
Please sign in to rate this answer.
Ahad Anjum 60 Reputation points

2024-04-02T15:55:59.16+00:00

@brtrach-MSFT thank you for the response! It turned out I had set a variable named PORT to be 1433 which was used in my connection string for SQL Server. I have changed the name of that environment variable and the application starting page works. Weirdly, the application does not seem to like when I add in my parameter to the URL (a crucial feature for the app), and it is causing the site to fail. In my Python code I grab it using:

email = request.args.get('email')

Not sure why this is happening. I will research and try to solve it. If you have any tips or advice on this, I'd greatly appreciate it. Thank you for your response!

brtrach-MSFT 15,251 Reputation points Microsoft Employee

2024-04-02T16:10:21.7733333+00:00

@Ahad Anjum Glad to hear that you were able to resolve the issue with the port number! As for the issue with the parameter in the URL causing the site to fail, it's possible that the parameter is not being passed correctly to the Flask app.

You mentioned that you are grabbing the parameter using request.args.get('email'). This should work if the parameter is passed in the URL as ?email=value. Make sure that the parameter name matches the one in your Flask app and that it's being passed correctly in the URL.

If the parameter is being passed correctly and you're still experiencing issues, you can try adding some logging statements to your Flask app to see where the issue is occurring. You can also try running the Flask app locally with the parameter to see if it works correctly.

Try the above and let me know the outcome by replying to this message.

A kind ask from you is to consider accepting my above response as an answer. This let's my leadership know we are providing helpful assistance to you. Thank you for your consideration.

Ahad Anjum 60 Reputation points

2024-04-02T18:27:45.18+00:00

@brtrach-MSFT The app does work locally, and I seem to be correctly passing the parameter. Here is more of the code for more clarity:

@app.route('/') def home(): global dataframes, memory, chat_prompt, context, userName, userImage, agent chat_prompt = chat_prompt_init() email = request.args.get('email') if not email: return render_template('no_email.html') dataframes = load_data_frames(email) context = dataframes.get("context") contractDetails = dataframes.get("contract_details") projectDetails = dataframes.get("project_details") storeDetails = dataframes.get("store_details") userName = dataframes.get("name") userImage = dataframes.get("profile_pic") memory = ConversationBufferWindowMemory() client = AzureChatOpenAI( api_key=os.getenv("AZURE_OPENAI_KEY"), api_version="2023-12-01-preview", azure_endpoint=os.getenv("AZURE_OPENAI_ENDPOINT"), deployment_name=os.getenv("DEPLOYMENT_NAME"), temperature=0 ) agent = create_pandas_dataframe_agent(llm=client, df=[storeDetails, contractDetails, projectDetails], verbose=True, agent_type=AgentType.OPENAI_FUNCTIONS, return_intermediate_steps=True) return render_template('Index.html')

I haven't set up any session state management so it can only run on one instance, I wonder if that is causing the issue. I am not sure though. I will add some logging to the app to help. Could it be the use of os.getenv() instead of os.environ[]?

brtrach-MSFT 15,251 Reputation points Microsoft Employee

2024-04-02T22:10:04.74+00:00

@Ahad Anjum It's good to hear that the parameter is being passed correctly and that the app works locally.

Regarding the issue with the parameter causing the site to fail, it's possible that the app is not handling the case where the parameter is not present correctly. In your code, you're checking if the email parameter is present using if not email: and rendering a different template if it's not present. However, if the email parameter is not present and you're rendering a different template, it's possible that the rest of the code in the home() function is not being executed correctly.

As for the session state management, if the app is running on multiple instances, each instance will have its own memory and dataframes, which could cause issues with the parameter being passed correctly. Can you verify how many instances your web app is currently running on?

Regarding the use of os.getenv() instead of os.environ[], both of these should work correctly for getting environment variables. However, os.environ[] is the recommended way to get environment variables in Python.

Adding logging statements to your Flask app is a good idea to help diagnose the issue. You can add logging statements using the logging module in Python. For example, you can add a logging statement like this:

import logging @app.route('/') def home(): logging.info('Home page requested') # rest of the code

This will log a message to the console when the home page is requested. You can add more logging statements throughout your code to help diagnose the issue.

Ahad Anjum 60 Reputation points

2024-04-03T14:48:44.97+00:00

@brtrach-MSFT I switched the os.getnenv to os.environ just in case. In terms of the app not handling the parameter correctly, the rest of the code in my home() function should only run if the email address parameter is present. I added logging to my requirements.txt file and to the app itself but now it is not deploying to Azure and giving me an error. Do you see anything I am missing? I have the import statement at the top of the file and even added a basicConfig.

@app.route('/') def home(): logging.basicConfig(level=logging.INFO) global dataframes, memory, chat_prompt, context, userName, userImage, agent chat_prompt = chat_prompt_init() logging.info('Home page requested') email = request.args.get('email') if not email: logging.info('email not found') return render_template('no_email.html') logging.info('email found') logging.info('loading dataframes') dataframes = load_data_frames(email) logging.info('df loaded') context = dataframes.get("context") contractDetails = dataframes.get("contract_details") projectDetails = dataframes.get("project_details") storeDetails = dataframes.get("store_details") userName = dataframes.get("name") userImage = dataframes.get("profile_pic") memory = ConversationBufferWindowMemory() logging.info('loading client') client = AzureChatOpenAI( api_key=os.environ["AZURE_OPENAI_KEY"], api_version="2023-12-01-preview", azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"], deployment_name=os.environ["DEPLOYMENT_NAME"], temperature=0 ) logging.info('client loaded') agent = create_pandas_dataframe_agent(llm=client, df=[storeDetails, contractDetails, projectDetails], verbose=True, agent_type=AgentType.OPENAI_FUNCTIONS, return_intermediate_steps=True) logging.info('agent made') return render_template('Index.html')

I am not sure if this is the correct way to check, but on the home page of the web app, the instance count is 1. I had two links of the web app open, and the number did not change. Maybe that is due to the plan I have which is Basic (B1)? I think if it is only running on one instance then there is no way for the parameter to work, right?

Ahad Anjum 60 Reputation points

2024-04-03T17:04:39.1966667+00:00

@brtrach-MSFT I got the logging to work. However, it does not show up in my log stream. Furthermore, I did some testing on hardcoding the email as an environment variable to load in my data from the SQL Query. Locally this worked so I tried it on the web app, and it does not work. Can't see the logs but I am assuming this means the issue could be my data base connection more so than grabbing the email parameter. Here is how I connect to the database before making queries:

from sqlalchemy import create_engine def load_data_frames(UserID: str): user = os.environ["USER_NAME"] password = os.environ["PASSWORD"] hostName = os.environ["HOSTNAME"] port = os.environ["SQLPORT"] db = os.environ["DATABASE"] # Establishing the database connection URI database_uri = f"mssql+pyodbc://{user}:{password}@{hostName}:{port}/{db}?driver=ODBC+Driver+17+for+SQL+Server" engine = create_engine(database_uri)

Let me know if you see anything wrong. I am more certain now that the issue lies with the database connection more than grabbing the email parameter after testing it hardcoded. Here is the home() function with the logging:

@app.route('/') def home(): global dataframes, memory, chat_prompt, context, userName, userImage, agent chat_prompt = chat_prompt_init() #email = request.args.get('email') email1 = os.environ["EMAIL1"] logging.info('email got') if not email1: return render_template('no_email.html') logging.info('email found') logging.info('loading dataframes') dataframes = load_data_frames(email1) logging.info('df found') context = dataframes.get("context") contractDetails = dataframes.get("contract_details") projectDetails = dataframes.get("project_details") storeDetails = dataframes.get("store_details") userName = dataframes.get("name") userImage = dataframes.get("profile_pic") memory = ConversationBufferWindowMemory() logging.info('client loading') client = AzureChatOpenAI( api_key=os.environ["AZURE_OPENAI_KEY"], api_version="2023-12-01-preview", azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"], deployment_name=os.environ["DEPLOYMENT_NAME"], temperature=0 ) logging.info('client made') agent = create_pandas_dataframe_agent(llm=client, df=[storeDetails, contractDetails, projectDetails], verbose=True, agent_type=AgentType.OPENAI_FUNCTIONS, return_intermediate_steps=True) return render_template('Index.html')

brtrach-MSFT 15,251 Reputation points Microsoft Employee

2024-04-04T00:43:51.65+00:00

@Ahad Anjum I'm glad to hear that you were able to get the logging to work. However, it's strange that the logs are not showing up in the log stream. You may want to check the log level of your Flask app to make sure that it's set to a level that will show the logs. You can also try checking the logs in the Azure portal by going to your App Service and selecting "Log stream" from the "Monitoring" section.

Regarding the issue with the database connection, the code you provided looks correct. However, it's possible that there is an issue with the connection string or the credentials. You may want to double-check that the credentials and connection string are correct, and that the SQL Server is accessible from the Azure App Service.

Using Kudu and tcpping to verify SQL connectivity:

Navigate to Kudu for your Azure web app. You can access it at https://<yourwebsitename>.scm.azurewebsites.net.

Open the CMD console located under the “Debug console” menu.

Run the tcpping command against your SQL server using the following syntax:
tcpping <servername>.database.windows.net:1433

If successful, you’ll see a response indicating that the connection is established.

If your instance count is listed as 1, then your request should be remaining on a single instance. I see that you're using containers. Are you using a multi container approach? If so, what are your containers configured as in terms of roles?

Ahad Anjum 60 Reputation points

2024-04-04T13:35:22.35+00:00

@brtrach-MSFT I am using a single container approach. I checked the log stream right now this morning and I was getting an error connecting to the SQL database. I also checked the Kudu and was not getting any response. I'll check with my team to see why this is happening. Thank you!

brtrach-MSFT 15,251 Reputation points Microsoft Employee

2024-04-07T17:55:14.43+00:00

@Ahad Anjum Sure, let me know their response so we can assist you further if needed.

Ahad Anjum 60 Reputation points

2024-04-17T14:06:29.3266667+00:00

@brtrach-MSFT It turns out that I needed to change the hostname for my SQL Server connection string to be the full name rather than the shortened version or the IP address. When I made this change, my web app restarted, and it was working! I did recently close the page and reload it and now I am getting 500 internal server errors. I haven't made any other changes to the code or the web app so I am wondering if this could be down time? Nothing shows up as an error in my log stream, everything seems fine there.
Sign in to comment

Azure Flask Web App Fails During Deployment (504 Gateway Timeout Error)

0 additional answers