Package: l2tpns
Version: 2.4.1-1
When working with a slow RADIUS server, l2tpns can lose track of IPs in
its assignable pool causing the daemon to eventually run out of
assignable addresses. The messages below show the last IP in the pool
being lost, and the next connection being rejected.
2025-09-17 11:29:36 03/00 New tunnel from 1.2.3.4:1701 ID 3
2025-09-17 11:29:40 03/14 Shutting down session 14: RADIUS timeout.
2025-09-17 11:29:40 03/14 Login by stats at 10.42.96.253 from 1.2.3.4
(host.example.com)
2025-09-17 11:29:55 03/14 Kill session 14 (stats): Expired
2025-09-17 11:29:56 03/00 Shutting down tunnel 3 (Stopped)
2025-09-17 11:31:06 03/00 Kill tunnel 3: Expired
2025-09-17 11:33:30 03/00 New tunnel from 1.2.3.4:1701 ID 3
2025-09-17 11:33:30 03/14 assign_ip_address(): out of addresses
2025-09-17 11:33:30 03/14 No IP allocated. The IP address pool is FULL!
2025-09-17 11:33:30 03/14 Shutting down session 14: No IP addresses
available.
2025-09-17 11:33:30 03/14 IPCP ConfigReq ignored in Dead phase
2025-09-17 11:33:45 03/14 Kill session 14 (stats): Expired
2025-09-17 11:33:50 03/00 Shutting down tunnel 3 (Stopped)
2025-09-17 11:35:00 03/00 Kill tunnel 3: Expired
Note that line 2 has a RADIUS timeout, and the subsequent line shows the
user successfully logging in. This appears to be the series of events
needed to trigger this bug.
I have built the current head of the l2tpns source, and it still
contains the same issue.
