pfSense » pfSense Plus

Netgate 6100 (8 GB RAM), pfSense Plus 26.07-RELEASE.

pfnet-controller resident memory grows continuously and does not stabilise.

Measurements on this install:
- 414 MB shortly after service start
- 646 MB after 3h35 uptime
- 2755 MB (34% of system RAM) after 24h16 uptime

Roughly 65 MB per hour, with no observed plateau. Restarting the service
reclaims all of it, which suggests a leak rather than a growing legitimate
working set.

Why this matters beyond memory usage:
CoreDNS runs inside pfnet-controller and binds port 53 on all internal
interfaces. Restarting the service to reclaim memory interrupts DNS for the
entire network for roughly ten seconds. On appliances with less RAM this
turns a slow leak into a recurring service interruption.

The service also proved fragile under load: issuing about twenty consecutive
feed enable/disable writes through the API caused pfnet-controller to stop
entirely, taking network DNS down until it was manually restarted. That may
or may not share a root cause with the memory growth.

Configuration on this install:
- 13 ThreatGate feeds (10 remote, 3 category/country)
- roughly 17000 IPv4 entries across the resulting pf tables
- CoreDNS with the rexdns plugin, three blocklists
- refresh_interval 24h0m0s

Related: #17029, #17030