Stack trace of crash with FR 3.0.10

[Alan DeKok]

On Nov 18, 2015, at 5:22 AM, Sebastian Hagedorn <Hagedorn at uni-koeln.de> wrote:
> 
> Hi,
> 
> we've been running FR 3.0.10 in production for about a month now. We have four production servers on RHEL 6. Every few days one of the radiusd processes quits without an error message. We have now enabled the panic action and installed the debuginfo RPMs. Last night we got the first usable gbd.log. Here's the relevant part:

  Thanks.
> 
> #22 0x000000396dc15dbc in fr_pair_value_strsteal (vp=0x7f8b48732dc0, src=0x7f8b4874f5e0 "leap:session-key=\200hl\240:l\236\255t\256\247\366Z\315%f\233\220q~\343-\207VΟ\334\366N%i\372p\274Q\242\243\232ީ/\257\362\315\376f\344\067>0\356\262\v\233\343\240\344pH\213\177") at /usr/src/debug/freeradius-server-3.0.10/src/lib/pair.c:2125
>       q = <value optimized out>

  LEAP?  Don't use that.  It's insecure.

> As far as we have been able to discern, previous crashes didn't all occur in post-proxy.

  OK.  So it's something else.

> The same user authenticated successfully before, so it also doesn't seem to related to that specific user. I will post future backtraces if they offer new evidence. Please let me know if additional info would be helpful.

  Any other back traces would help.

  This might be hard to track down without local tests...

  Alan DeKok.