Reply the requests logged in NGINX access.log and error.log files, with the help of some utility, to degub API crashing issue.
In a situation in which few logs were available to debug a crash of API component (only NGINX logs, and few more), automatically parse available logs to reply the requests and to identify the "bad ones".