Craig Campbell wrote:
I have attached the radius.log file below, as well as gdb sessions for the hung processes showing the results of the gd 'bt' and 'list' commands.
The log is interesting.
Sat Oct 17 02:01:25 2009 : Error: WARNING: Unresponsive child for request 165616, in module sql component accounting
It looks like the SQL module is being blocked somehow. Why?
(gdb) bt #0 0x0000003acf4dee6e in __lll_lock_wait_private () from /lib64/libc.so.6 #1 0x0000003acf48c75d in _L_lock_1685 () from /lib64/libc.so.6 #2 0x0000003acf48c4a7 in __tz_convert () from /lib64/libc.so.6 #3 0x00002b794fa6e39f in vp_prints_value (out=0x41ced6c0 "", outlen=1008, vp=<value optimized out>, delimitst=<value optimized out>) at print.c:267
Hmm... OK. It's a pthread lock / fork issue. I think the solution is to print the "env" variables *before* forking. I'll see what I can do. Alan DeKok.