Apache Reverse Proxy saturation due to NATS/Mesh WebSocket connections

[polidela]

Environment

OS: Debian
Proxy: Apache/2.4.66

I am running Tactical RMM behind a corporate Apache Reverse Proxy (Debian based, Apache 2.4.66). I am experiencing a severe resource saturation issue where Apache fills its scoreboard, eventually blocking not only the RMM but all other virtual hosts configured on the same proxy.
Here the log
[Mon Feb 02 13:16:59.736528 2026] [mpm_event:error] [pid 723:tid 723] AH03490: scoreboard is full, not at MaxRequestWorkers.Increase ServerLimit.

Questions:
Are there specific recommended apache or mpm_event tuning parameters (ServerLimit, ThreadsPerChild, AsyncRequestWorkerFactor) specifically tested for large agent deployments behind Apache?

Is there a known issue with how Apache handles the NATS WebSocket heartbeat compared to the native Nginx configuration provided in the docs?

[PC-Support-Services]

This is what claude has to say on the matter:

https://claude.ai/share/5219e73e-8082-42b3-9e99-2b942be13d1f