rlm_jradius linking problems
Hi there, Point me in the right direction if I'm talking to the wrong people here :). I'm trying to get JRadius up and running, but not having much success. I've built rlm_jradius, but for some reason it is not creating .so objects for freeradius to load. As a result its crashing out with this error: radiusd.conf[1515] Failed to link to module 'rlm_jradius': /usr/local/radius/lib/rlm_jradius.a: invalid ELF header Has anyone had any success with JRadius, and if so can they possibly help me? Thanks all. -- BigMick
I have implemented a file scanning mechanism to scan the radius accounting detail file and subsequently upload to database server but at the time of scanning, I detect the presence of a "yesterday" file ( ie a completed file). This will mean that my accounting record inside the database is one day late. Now I understand there is a way to instruct radius server to change the file name hourly, so theoretically I should be able to scan the presence of "last hour" completed file, and then upload to database server. However, assumming the scanning, processing, and subsequent uploading to database server is very slow, it could mean that from the start of one scan to the next scan, if more than one hour has passed, I would have missed one of the "last hour" file. Anyone has a better idea of how to process an hourly file more gracefully ?
Ming-Ching Tiew wrote:
I have implemented a file scanning mechanism to scan the radius accounting detail file and subsequently upload to database server but at the time of scanning, I detect the presence of a "yesterday" file ( ie a completed file). This will mean that my accounting record inside the database is one day late.
Now I understand there is a way to instruct radius server to change the file name hourly, so theoretically I should be able to scan the presence of "last hour" completed file, and then upload to database server.
However, assumming the scanning, processing, and subsequent uploading to database server is very slow, it could mean that from the start of one scan to the next scan, if more than one hour has passed, I would have missed one of the "last hour" file.
Anyone has a better idea of how to process an hourly file more gracefully ? This is probably a stupid question but whay not log the accounting directly to the sql via the sql module?
-- Lewis Bergman Texas Communications 4309 Maple St. Abilene, TX 79602-8044 Off. 325-691-1301 Cell 325-439-0533 fax 325-695-6841
From: "Lewis Bergman" <lbergman@wtxs.net>
This is probably a stupid question but whay not log the accounting directly to the sql via the sql module?
Reasons :- I am logging to MSSQL and I have tried in the past to do it directly, I find that the stability is POOR and reliability is NOT ACCEPTABLE. For example, the sql driver does not reconnect upon failure. I tried fixing it myself but I also faced other weird problems which are difficult to troubleshoot. Also per record logging is way too slow to cope with the speed I am looking for. I am in fact using the TDS BULK INSERT mechanism for insertion. It is much much faster than per record logging. All in all, I find that logging to database via radius server directly is a bad design, more so for a heavy radius server with lots of traffic, and worse if it is across the network (WAN!) , due to network instablity, database server load conditions etc etc etc. Cheers
"Ming-Ching Tiew" <mingching.tiew@redtone.com> wrote:
I am logging to MSSQL and I have tried in the past to do it directly, I find that the stability is POOR and reliability is NOT ACCEPTABLE. For example, the sql driver does not reconnect upon failure. I tried fixing it myself but I also faced other weird problems which are difficult to troubleshoot.
See rlm_sql_log in the 1.1.0-pre0 image. It should help. Alan DeKok.
Actually I was told by the development of such thing ( ie decoupled SQL logging ) in the radius server some time ago, that is a good thing but I am currently using my own relay logging and it is already very stable and very fast ( using bulk insert ), it just suffers the limitation that it is one day late, and thus I have these questions :- The radius server method, I believe is also based on scanning a directory of files, how does it handle files which are still growing ( ie unfinished files ) ? Or it is assuming that the files have been completed ( ie there are no files which are still active ! ) ? Cheers ----- Original Message ----- From: "Alan DeKok" <aland@ox.org> To: "FreeRadius users mailing list" <freeradius-users@lists.freeradius.org> Sent: Saturday, December 24, 2005 1:38 AM Subject: Re: Radius accounting file scanning and upload to database
"Ming-Ching Tiew" <mingching.tiew@redtone.com> wrote:
I am logging to MSSQL and I have tried in the past to do it directly, I find that the stability is POOR and reliability is NOT ACCEPTABLE. For example, the sql driver does not reconnect upon failure. I tried fixing it myself but I also faced other weird problems which are difficult to troubleshoot.
See rlm_sql_log in the 1.1.0-pre0 image. It should help.
Alan DeKok. - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
"Ming-Ching Tiew" <mingching.tiew@redtone.com> wrote:
The radius server method, I believe is also based on scanning a directory
No.
of files, how does it handle files which are still growing ( ie unfinished files ) ? Or it is assuming that the files have been completed ( ie there are no files which are still active ! ) ?
No. The server & "radsqlrelay" program co-operate with locking to ensure that growing files are handled. Alan DeKok.
participants (4)
-
Alan DeKok -
BigMick -
Lewis Bergman -
Ming-Ching Tiew