A little update... In the same time (2009-02-17 11:23:16), there was at least 20 other closed sessions and they are from different NASes so I guess that problem is on the server side... In which cases accounting_onoff_query is triggered by radius server? (Accounting Off, and Stop but is there any other ...) I partially solved this problem with cron script which reopens the closed sessions but that is not a long term solution :) Marinko Tarlac wrote:
I have at Tue Feb 17 11:22:44 2009 and Tue Feb 17 11:24:14 2009 which is OK because the updates are set to every 90 seconds.
According to database the session was closed at 2009-02-17 11:23:16
After Tue Feb 17 11:24:14 2009, radius still receives regular updates. acctterminatecause is empty and when I kill this session in NAS list, I see that Stop was received and the same session is updated (acctterminatecause and AcctStopTime was updated which means that NAS didn't lost this session and still has its unique ID.)