Increment Acct-Delay-Time on proxy
Hello, I've installed FreeRADIUS 2.1.7 on a CentOS system. In my setup, the RADIUS server will work as proxy, storing locally the accounting packets if the destination servers are down, and resending them when the destination servers are up again. To achieve this result I've followed the "robust-proxy-accounting" example config file with success. Now I'm looking for a way to update the "Acct-Delay-Time", so that, when the destination servers are back, the packet is read from the "detail" file and is sent by the proxy adding to the Acct-Delay-Time the delay introduced by the temporary failure at the proxy level. Is there a way to do this with FreeRADIUS? Thanks! Emanuele
Emanuele Colombo wrote:
I've installed FreeRADIUS 2.1.7 on a CentOS system. In my setup, the RADIUS server will work as proxy, storing locally the accounting packets if the destination servers are down, and resending them when the destination servers are up again. To achieve this result I've followed the "robust-proxy-accounting" example config file with success.
Now I'm looking for a way to update the "Acct-Delay-Time", so that, when the destination servers are back, the packet is read from the "detail" file and is sent by the proxy adding to the Acct-Delay-Time the delay introduced by the temporary failure at the proxy level. Is there a way to do this with FreeRADIUS?
The detail file reader will automatically do that for you. Alan DeKok.
----- Messaggio originale -----
Da: "Alan DeKok" <aland@deployingradius.com> A: "FreeRadius users mailing list" <freeradius-users@lists.freeradius.org> Inviato: Lunedì, 13 febbraio 2012 17:54:01 Oggetto: Re: Increment Acct-Delay-Time on proxy
Now I'm looking for a way to update the "Acct-Delay-Time", so that, when the destination servers are back, the packet is read from the "detail" file and is sent by the proxy adding to the Acct-Delay-Time the delay introduced by the temporary failure at the proxy level. Is there a way to do this with FreeRADIUS?
The detail file reader will automatically do that for you.
Actually it didn't worked for me... How can I check this? Is there any setting for enabling this feature or a delay threshold after it starts working? What configuration file I could check? Thanks! Emanuele
Emanuele Colombo wrote:
Actually it didn't worked for me... How can I check this? Is there any setting for enabling this feature or a delay threshold after it starts working? What configuration file I could check?
The code has been in the server since August 2007, i.e. 2.0.0. There is no setting for enabling it. It works. What do you mean "it didn't work for me" ? Alan DeKok.
----- Messaggio originale -----
Da: "Alan DeKok" <aland@deployingradius.com> A: "FreeRadius users mailing list" <freeradius-users@lists.freeradius.org> Inviato: Lunedì, 13 febbraio 2012 18:03:42 Oggetto: Re: Increment Acct-Delay-Time on proxy
The code has been in the server since August 2007, i.e. 2.0.0.
There is no setting for enabling it. It works.
What do you mean "it didn't work for me" ?
I looked at packets sent by my FreeRADIUS server when the destination was back and Acct-Delay-Time was 0, as the original ones. Now I know that it should already work so tomorrow I'll make some other test and I'll try to find where my mistake is. Thanks, Emanuele
----- Messaggio originale -----
Da: "Alan DeKok" <aland@deployingradius.com> A: "FreeRadius users mailing list" <freeradius-users@lists.freeradius.org> Inviato: Lunedì, 13 febbraio 2012 18:03:42 Oggetto: Re: Increment Acct-Delay-Time on proxy
The code has been in the server since August 2007, i.e. 2.0.0.
There is no setting for enabling it. It works.
What do you mean "it didn't work for me" ?
Hi Alan, I've made some tests today and it works like you said. I think I made an error yesterday while testing, sorry... I've also updated to the latest release (2.1.12), which solved some other problems. Actually I'm experiencing the problem described here, or something similar: https://bugs.freeradius.org/bugzilla/show_bug.cgi?id=128 As it should already be fixed in 2.1.12, I'll make some more tests tomorrow, but the behaviour of my proxy is: - if I use a single detail file it works without problems in modules/details.mytest I have: detailfile = ${radacctdir}/detail.mytest/detail-mytest in sites-availabe/mytest I have: filename = "${radacctdir}/detail.mytest/detail-mytest" - if I use the "dynamic" file name (as in robust-proxy-accounting provided file) it continuously adds lines like: Packet-Original-Timestamp = "Feb 14 2012 14:40:24 CET" Proxy-State = 0x3231393933 in modules/details.mytest I have: detailfile = ${radacctdir}/detail.mytest/detail-%Y%m%d:%H in sites-availabe/mytest I have: filename = "${radacctdir}/detail.mytest/detail-*:*" As I said, tomorrow I'll make some other tests, but if you have any hint about this behaviour let me know! ;-) Thanks for you help! Emanuele
Emanuele Colombo wrote:
Hi Alan, I've made some tests today and it works like you said. I think I made an error yesterday while testing, sorry... I've also updated to the latest release (2.1.12), which solved some other problems.
That's good.
Actually I'm experiencing the problem described here, or something similar: https://bugs.freeradius.org/bugzilla/show_bug.cgi?id=128
As it should already be fixed in 2.1.12, I'll make some more tests tomorrow, but the behaviour of my proxy is:
- if I use a single detail file it works without problems in modules/details.mytest I have: detailfile = ${radacctdir}/detail.mytest/detail-mytest in sites-availabe/mytest I have: filename = "${radacctdir}/detail.mytest/detail-mytest"
- if I use the "dynamic" file name (as in robust-proxy-accounting provided file) it continuously adds lines like: Packet-Original-Timestamp = "Feb 14 2012 14:40:24 CET" Proxy-State = 0x3231393933 in modules/details.mytest I have: detailfile = ${radacctdir}/detail.mytest/detail-%Y%m%d:%H in sites-availabe/mytest I have: filename = "${radacctdir}/detail.mytest/detail-*:*"
Maybe the globbing isn't done right... it's hard to test.
As I said, tomorrow I'll make some other tests, but if you have any hint about this behaviour let me know! ;-)
Not for now. Alan DeKok.
participants (2)
-
Alan DeKok -
Emanuele Colombo