Man, status code 499 is the worst! Had this issue last month with our API. Turns out the client-side timeout was set to 2 seconds, but the server took 3+ seconds to process heavy requests.
We bumped the timeout to 5s and added retry logic. Also, used New Relic to track slow endpoints. Fixed most of it!
If you're on Nginx, check `proxy_ignore_client_abort on;`—it *might* help, but be careful with it.
We bumped the timeout to 5s and added retry logic. Also, used New Relic to track slow endpoints. Fixed most of it!
If you're on Nginx, check `proxy_ignore_client_abort on;`—it *might* help, but be careful with it.