a month ago
Our users are seeing a lot of 502s lately, seemingly after new deployments.
The strange thing is we also aren't seeing any 502s in our HTTP logs (I've been looking at threads similar to https://station.railway.com/questions/intermittent-502-errors-535c0db3).
Some theories here:
- Clients are making requests for non-existent / old assets from previous deploy (totally possible)
- There is some Railway DNS / proxy issue that is upstream of our app that we dont have good visibility into.
Either way, we're lacking visibility and wanted to see if there's anything weird your team can observe beyond what we can see.
Thanks.
2 Replies
Status changed to Awaiting Railway Response Railway • about 1 month ago
a month ago
This thread has been marked as public for community involvement, as it does not contain any sensitive or personal information. Any further activity in this thread will be visible to everyone.
Status changed to Open Railway • about 1 month ago
a month ago
Hello imsteev,
two things worth trying , set RAILWAY_DEPLOYMENT_OVERLAP_SECONDS and RAILWAY_DEPLOYMENT_DRAINING_SECONDS, draining defaults to 0 so in-flight requests get cut the moment a new deploy goes live also whenever you catch a 502 , grab the X-Railway-Request-Id from the response header and share it with railway support, that's the only way they can trace it on their end
Hope this help you :)
a month ago
To prevent downtime during new deployments, configure a healthcheck path so traffic only routes to your app once it has fully started. For more control over your deployments you can check Deployment Teardown.