"
Good day,
when using Captive Portal with 1400-1700 users and 1.5-3 Gbit/s traffic, packet loss and RTT (ping) increase periodically occur. At the same time, the average CPU load increases to 45. When disabling Captive Portal, everything stabilizes, the average CPU load does not rise above 5. OPNsense is used on 2 identical servers (hot standby using CARP), the product versions used are OPNsense 24.1.10_8-amd64 and OPNsense 25.7.2-amd64.
With OPNsense 24.1.10_8-amd64, there are no problems when enabling Captive Portal and the average CPU load does not rise above 7 with a similar number of users and traffic.
Server specs: Intel Xeon Gold 6252 x2, 384GB RAM, Mellanox CX4121A 25Gbit 2-port network card.
When analyzing the problem, I noticed that in the top by CPU usage (by time and quantity) python3.11 (/usr/local/opnsense/scripts/OPNsense/CaptivePortal/cp-background-process.py) and lighttpd.
Maybe there is a solution to this problem?
when using Captive Portal with 1400-1700 users and 1.5-3 Gbit/s traffic, packet loss and RTT (ping) increase periodically occur. At the same time, the average CPU load increases to 45. When disabling Captive Portal, everything stabilizes, the average CPU load does not rise above 5. OPNsense is used on 2 identical servers (hot standby using CARP), the product versions used are OPNsense 24.1.10_8-amd64 and OPNsense 25.7.2-amd64.
With OPNsense 24.1.10_8-amd64, there are no problems when enabling Captive Portal and the average CPU load does not rise above 7 with a similar number of users and traffic.
Server specs: Intel Xeon Gold 6252 x2, 384GB RAM, Mellanox CX4121A 25Gbit 2-port network card.
When analyzing the problem, I noticed that in the top by CPU usage (by time and quantity) python3.11 (/usr/local/opnsense/scripts/OPNsense/CaptivePortal/cp-background-process.py) and lighttpd.
Maybe there is a solution to this problem?
