Hey there! We've found the following might help you get unblocked faster:

• 🧵 Flask App + Websocket Server Complications

• 🧵 API intermittently unavailable – “Error: getaddrinfo ENOTFOUND”

• 🧵 Streamlit 502 Bad Gateway Error

If you find the answer from one of these, please let us know by solving the thread!

Naive response:

Are you sure the recent error isn't related to your application crashing?

I see:

raceback (most recent call last):

File "/app/Leadoff/utils/cron.py", line 998, in background_task

refresh_crm_data()

File "/app/Leadoff/utils/cron.py", line 964, in refresh_crm_data

opportunity_update["amount"] = int(deal_amount)

ValueError: invalid literal for int() with base 10: '2539.8'

Interesting - I've seen errors like this but the server continues to run - is there a certain number of errors that would trigger some sort of shutdown? Errors will occur, so just wondering why this one would be "instance killing". Is there a better way to be handling errors that occur to not spin down the server.

Also - if this was the reason, why isn't it clear in the logs? I simply see our Cron Job heartbeat print statements and then nothing passed a certain time.

Thank you!

I wouldn't know, the issue with OOMs/crashes is that you are never guaranteed a graceful shutdown. The only thing that we can do from the Railway side is see if:

1. We propagate the error correctly
2. We are logging

When I review our dashboards, nothing is telling me that we're dropping the ball here. (And believe me, we've dropped the ball.) Sorry I don't have much more for you here.

Appreciate the help - any idea why our health check would be intermittently getting a timeout? This seems really peculiar. I've looked at our usage statistics and it doesn't seem like we are anywhere near a max capacity that would cause delays.

I took a look at the logs for your last deployment from yesterday, and it doesn't have any 499s. Did you change anything in your app that could have addressed this?