[Segmentation Fault while proxing Request to home server]

[Thomas Fagart]

Hello,

We're having the following authentication scenario for Wimax User 
Authentication :


ASN --  EAP/TTLS ---> 2 freeradius used as proxy --> 2 freeradius home 
server.

AAA is fine when both home servers are up,but when one of them or both 
of them are dead, we're having the following logs on the proxy :


rad_recv: Access-Request packet from host a.b.c.D port 10008, id=117, 
length=406
Sending duplicate proxied request to home server a.b.c.d port 1812 - ID: 36
Sending Access-Request of id 36 to a.b.c.d port 1812
         User-Name = "xx at domain.com"
         MS-CHAP-Challenge = 0xdcd70de41d9783aa76aa573d3d07f84d
         MS-CHAP2-Response = 
0x4700db77a532a5ec9b28c3805d18cc35ed170000000000000000d00f9d99d46d688436477daf0cdd734b813f11e4bd115e16
         NAS-IP-Address = x.y.z.w        Calling-Station-Id = "002104bec153"
         NAS-Identifier = "WASN"
         Event-Timestamp = "Nov 18 2010 01:43:06 CET"
         WiMAX-Release = "1.1"
         WiMAX-Capability = 0x0105312e31020302030301040301
         WiMAX-Accounting-Capabilities = Flow-Based
         WiMAX-Hotlining-Capabilities = Hotline-Profile-Id
         WiMAX-Idle-Mode-Notification-Cap = Supported
         WiMAX-Attr-1281 = 0x01
         WiMAX-BS-Id = 0x303030303261323930633030
         WiMAX-GMT-Timezone-offset = 3600
         NAS-Port-Type = Wireless-802.16
         WiMAX-Available-In-Client = 99
         WiMAX-PPAC = 0x010600000063
         Service-Type = Framed-User
         Message-Authenticator := 0x00000000000000000000000000000000
         Proxy-State = 0x313137
Waking up in 21.0 seconds.
WARNING: Internal sanity check failed in event handler for request 6: 
Discarding the request!
Segmentation fault

And then the proxy radius dies.

This happens each time at the fifth try while trying to send the request 
to a dead home server.

Regular logs gives that
Nov 18 01:37:46 vma-prdaut-08 radiusd[11028]: Marking home server 
a.b.c.d port 1812 as zombie (it looks like it is dead).


I know I should give a gdb trace to help, but since this is production 
server, it might take some times to give the trace.

Do you have any idea to what it could be related ?

Environment is the following :
FreeBSD 8.0/ Freeradius 2.1.10 compiled on threaded perl for snmp script 
use.

Thanks

Thomas