If a server crashes, we should send the last N logs of the server no matter what since that usually contains the crash info.
The following is a problematic scenario that can take place with the current lobby's log throttler system:
Malicious user starts attempting to attack server
Connections spam and cause unexpectedly high amounts of logs printed
Logs get ratelimitted
Malicious user successfully crashes the server but stacktrace and final debug information is never shown due to the fact the logs are being truncated by the ratelimitter
Alternatives solutions
Solution would be to strictly save the last 50-100 logs no matter what. If the ratelimit happens to be hit, then only begin deleting logs after they get old aka only delete it's not in the last 50 lines.
Description
If a server crashes, we should send the last N logs of the server no matter what since that usually contains the crash info.
The following is a problematic scenario that can take place with the current lobby's log throttler system:
Alternatives solutions
Solution would be to strictly save the last 50-100 logs no matter what. If the ratelimit happens to be hit, then only begin deleting logs after they get old aka only delete it's not in the last 50 lines.
Prior art