That should happen only if IP allocation has expired (see lease-duration in sqlippool.conf). There is another allocate-find query that issues random IPs.
Hmmm, maybe there is another problem in my config. I tried two requests within ten seconds. Attached you'll find the debug. During the second request the first ip-address is freed and can be used again. The lease-duration has the standard value of 3600, so this can't be the reason.
This is the table radippool after the second request:
+-----------+-----------------+--------------+---------------------+----------+----------+ | pool_name | framedipaddress | nasipaddress | expiry_time | username | pool_key | +-----------+-----------------+--------------+---------------------+----------+----------+ | poolUK | 10.10.10.10 | 10.98.6.95 | 2009-02-02 10:14:32 | peter2 | | | poolUK | 10.10.10.11 | | 2009-02-02 09:14:31 | | 0 | +-----------+-----------------+--------------+---------------------+----------+----------+
You don't have a pool_key because you are doing radtest requests. Proper request will have NAS-Port or Calling-Station-Id as pool_key. With updated queries user, nas *and* pool_key need to match for IP to be released. Queries in the distribution don't have pool_key so double login will release the older IP. Ivan Kalik Kalik Informatika ISP