[Kea-users] statistics-get - computation of # of current leases per pool?

[Václav Ovsík]

Fellow users of KEA,

I'm trying to migrate from ISC dhcpd to KEA. The current dhcpd server
installations are working in HA load balancing mode and I decided to
use HA hot-standby mode for KEA (according to documentation).
The servers are running Debian 13 Trixie and KEA from trixie-backports version
3.0.3-1~bpo13+1.

Along the KEA replacement I also install monitoring plugin for KEA in
CheckMK (https://codeberg.org/mirabilos/kea-dhcp).  It monitors pools
usage by asking the KEA, but I have problem with the results.

My KEA instance has many static host reservations and some pools in some
subnets.

Specifically is problematic subnet with definition:

        {
          "id": 80,
          "subnet": "10.8.160.128/25",
          "option-data": [
            { "name": "routers", "data": "10.8.160.129" },
          ],
          "pools": [ { "pool": "10.8.160.242 - 10.8.160.250" } ]
        },

There is no dynamic lease from pool, but more static reservations outside of
pool, but the monitoring plugins show the problem - the pool is full because:

  # echo '{ "command": "statistic-get", "arguments": { "name": "subnet[80].pool[0].assigned-addresses" } }' | socat UNIX-CONNECT:/run/kea/kea4-ctrl-socket STDIO | json_pp
  {
     "arguments" : {
        "subnet[80].pool[0].assigned-addresses" : [
           [
              9,
              "2026-10-29 08:17:47.707752"
           ],
           [
              0,
              "2026-10-29 08:17:47.707562"
           ]
        ]
     },
     "result" : 0
  }

Yes, I have 9 leases, but outside of pool. Is this bug in the KEA server?

  # socat /run/kea/kea4-ctrl-socket STDIO <<< '{ "command": "lease4-get-page", "arguments": { "from": "start", "limit": 100 } }' | jq  -r '.arguments.leases[] | select(."subnet-id" == 80)."ip-address"' 
  10.8.160.140
  10.8.160.141
  10.8.160.143
  10.8.160.145
  10.8.160.146
  10.8.160.147
  10.8.160.148
  10.8.160.150
  10.8.160.151

The values in the level subnet[80] and subnet[80].pool[0] are the same.

  nssrv1.lim.i.cz:/etc/kea# echo '{ "command": "statistic-get", "arguments": { "name": "subnet[80].assigned-addresses" } }' | socat UNIX-CONNECT:/run/kea/kea4-ctrl-socket STDIO | json_pp
  {
     "arguments" : {
        "subnet[80].assigned-addresses" : [
           [
              9,
              "2026-10-29 08:17:47.707724"
           ],
           [
              0,
              "2026-10-29 08:17:47.707553"
           ]
        ]
     },
     "result" : 0
  }

And especially confusing is the value 0 and 9 in very close point in time.

Server restart doesn't help:

  # systemctl restart kea-dhcp4-server.service 
  # echo '{ "command": "statistic-get", "arguments": { "name": "subnet[80].assigned-addresses" } }' | socat UNIX-CONNECT:/run/kea/kea4-ctrl-socket STDIO | json_pp
  {
     "arguments" : {
        "subnet[80].assigned-addresses" : [
           [
              9,
              "2026-10-29 18:27:52.139910"
           ],
           [
              0,
              "2026-10-29 18:27:52.139706"
           ]
        ]
     },
     "result" : 0
  }

I think the value is simply bad according to
https://kea.readthedocs.io/en/kea-3.0.3/arm/dhcp4-srv.html#dhcp4-stats

Sorry for my English.
Any idea?
Regards
-- 
Zito