Aborted when seeing Failing proxied request

[Lai Fu Keung]

With the new version 3.0.8 running on our environment, I would say the problem is just slightly improved. Freeradius crashes from "several times an hour" in v3.0.7 to "once every several hours" in v3.0.8, after seeing "Proxy: Marking home server x.x.x.x as zombie (it has not responded in 30.000000 seconds)" message.

We have more than 10 radius home servers configured. Some of them are quite busy servers and I understand that sometimes they may respond slowly or even do not respond to the proxied requests. I am not familiar with the logic inside freeradius. But can freeradius at least not to abort itself when handling proxy servers with no response. 

Sorry that I can't configure our production home servers in our testing environment to debug freeradius. But anything else I can do??

Fu

-----Original Message-----
From: Freeradius-Users [mailto:freeradius-users-bounces+tfklai=hku.hk at lists.freeradius.org] On Behalf Of Alan DeKok
Sent: Wednesday, March 11, 2015 10:16 PM
To: FreeRadius users mailing list
Subject: Re: Aborted when seeing Failing proxied request

On Mar 11, 2015, at 3:45 AM, RESTOUX, Loïc <loic.restoux at capgemini.com> wrote:
> During a test, if I stop the home server, I'm getting "Failing proxied request" errors, then FreeRADIUS crashes. No log about zombie, though.
> 
> Backtrace:
> #0  0x0000003535a0f3cb in raise () from /lib64/libpthread.so.0
> #1  0x00007f6a00d216c6 in skgesigOSCrash () from /opt/oracle/instantclient_11_2/libclntsh.so.11.1
> #2  0x00007f6a00fd2f79 in kpeDbgSignalHandler () from /opt/oracle/instantclient_11_2/libclntsh.so.11.1
> #3  0x00007f6a00d218d6 in skgesig_sigactionHandler () from /opt/oracle/instantclient_11_2/libclntsh.so.11.1
> #4  <signal handler called>

  Uh… that’s bad.  The Oracle client library grabs the signal handler for itself?  It’s not the cause of the crash, but it’s terrible practice.

> #5  0x00007f6a029a0652 in fr_event_delete (el=0x19747f0, parent=0x66dd280) at /home/admwpsrs/src/freeradius-server-3.0.x-c10/src/lib/event.c:192
> #6  0x00007f6a029a0766 in fr_event_run (el=0x19747f0, when=<value optimized out>) at /home/admwpsrs/src/freeradius-server-3.0.x-c10/src/lib/event.c:308
> #7  0x00007f6a029a0a4d in fr_event_loop (el=0x19747f0) at /home/admwpsrs/src/freeradius-server-3.0.x-c10/src/lib/event.c:625
> #8  0x000000000042598c in main (argc=3, argv=<value optimized out>) at /home/admwpsrs/src/freeradius-server-3.0.x-c10/src/main/radiusd.c:581

  That’s the problem we’ve been fighting for a while.  Unfortunately, I can’t reproduce it here.

  What does the rest of the debug log say?  Knowing that might also help.  You could also try turning on DEBUG_STATE_MACHINE in src/main/process.c.  It will cause the debug output to print out state transitions for each request.

  Alan DeKok.


-
List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html