Re: Duplicate Acct-Status packets
Ivan, its interesting. Our NAS is linux pppoe-server with pppd last version. Pppd retransmits Acct-Status in 10 seconds. Is it possible to increase freeradius keep of tracking time, you talking about, from 5 to 15 seconds? Where is this place in configs/sources? See cleanup_delay in radiusd.conf. Ivan Kalik Kalik Informatika ISP - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
Duplicate Acct-Status packets
I'm running freeradius-1.1.7: NAS server sends Acct-Status-Start packet, then freeradius processes it, and confirms with Acct-Status-Reply. But if the reply packet is lost due UDP-nature, NAS sends second Acct-Status-Start. In this situation I have two duplicate records in radacct table. How can I avoid this? Unique db-indexes, or better something else? - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
Re: Duplicate Acct-Status packets
Unique will solve your problem but keep in mind that this will consume more resources. On Tue, Mar 31, 2009 at 11:50 AM, Bogomolov Andrei d...@clink.ru wrote: I'm running freeradius-1.1.7: NAS server sends Acct-Status-Start packet, then freeradius processes it, and confirms with Acct-Status-Reply. But if the reply packet is lost due UDP-nature, NAS sends second Acct-Status-Start. In this situation I have two duplicate records in radacct table. How can I avoid this? Unique db-indexes, or better something else? - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
Re: Duplicate Acct-Status packets
Unique will solve your problem but keep in mind that this will consume more resources. That is not a very good solution. True, there will be only one insert in the table but if you make sql fail the insert, sql module will fail as will accounting, so no response will go back to the NAS. And it will send that accounting packet again and again and again ... Under normal circumstances one dropped response is not a problem. NAS tends to repeat unanswered accounting packets every 2 seconds, while freeradius keep the request on the list for 5 seconds. So you would need three consecutive dropped responses for duplicate to end up in the radacct table. If your network has such packet loss fiddling with accounting is least of your worries. But if if your NAS is retransmitting in intervals longer than 5 seconds you should fix that. Ivan Kalik Kalik Informatika ISP On Tue, Mar 31, 2009 at 11:50 AM, Bogomolov Andrei d...@clink.ru wrote: I'm running freeradius-1.1.7: NAS server sends Acct-Status-Start packet, then freeradius processes it, and confirms with Acct-Status-Reply. But if the reply packet is lost due UDP-nature, NAS sends second Acct-Status-Start. In this situation I have two duplicate records in radacct table. How can I avoid this? Unique db-indexes, or better something else? - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
Re: Duplicate Acct-Status packets
Ivan, its interesting. Our NAS is linux pppoe-server with pppd last version. Pppd retransmits Acct-Status in 10 seconds. Is it possible to increase freeradius keep of tracking time, you talking about, from 5 to 15 seconds? Where is this place in configs/sources? t...@kalik.net wrote: Unique will solve your problem but keep in mind that this will consume more resources. That is not a very good solution. True, there will be only one insert in the table but if you make sql fail the insert, sql module will fail as will accounting, so no response will go back to the NAS. And it will send that accounting packet again and again and again ... Under normal circumstances one dropped response is not a problem. NAS tends to repeat unanswered accounting packets every 2 seconds, while freeradius keep the request on the list for 5 seconds. So you would need three consecutive dropped responses for duplicate to end up in the radacct table. If your network has such packet loss fiddling with accounting is least of your worries. But if if your NAS is retransmitting in intervals longer than 5 seconds you should fix that. Ivan Kalik Kalik Informatika ISP On Tue, Mar 31, 2009 at 11:50 AM, Bogomolov Andrei d...@clink.ru wrote: I'm running freeradius-1.1.7: NAS server sends Acct-Status-Start packet, then freeradius processes it, and confirms with Acct-Status-Reply. But if the reply packet is lost due UDP-nature, NAS sends second Acct-Status-Start. In this situation I have two duplicate records in radacct table. How can I avoid this? Unique db-indexes, or better something else? - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
