upstream connect error or disconnect/reset before headers.
One of my older/extremely low traffic apps as a pet project is having interimettent issues where this happens:
7 Replies
Project ID:
0c213cbd-4233-4cb7-9ef1-615c05cfe3650c213cbd-4233-4cb7-9ef1-615c05cfe365
You're free to look at the project and try to get the error: https://runnerspace.xevion.dev/
No specific page. Seems like an issue with kubernetes or whatever platform you're using to manage all the docker instances.
Happens on mobile too. Just drops the connection. I think it's seizing up on the CSS too.
looks like Railway may be having an incident. other users are reporting this too
yea we're seeing this too, flakily, across all of our apps
I notified the team, they’ll be handling this shortly
🚀 Cool. Didn't see any other reports yet so I was worried I was gonna have to start debugging a really old project 😭
looking into it
We isolated the issue to resource exhaustion on some us-east proxies. We've retroactively created an incident with this context.
Basically AWS SNS started hammering our proxies with a large number of connections solely in that region and it exhausted connections. We've shipped a mitigation, and also are adding more capacity to that region.
Traffic ingested from other edges (us-west/europe/asia) were unaffected.
https://railway.instatus.com/cln27mrdl75373bbofzz3mh17l