I am running OSRM version 5.24.0, on Debian 10.9, AWS EC2 m5.xlarge with 4 vCPUs and 16GB of RAM, CH algorithm, map that is combination of France, Belgium, Nederlands... Keep alive connections are enabled.
Memory usage on the server is 40% by osrm-routed, the rest is cached files.
CPU is usually around 20%, all queries are resolved in milliseconds.
Then out of nowhere, CPU is on 100%, response time increases. I restart osrm-backend and everything goes back to normal.
Did anyone face the similar problem or have any idea how to try to pinpoint the issue?