Unusual pstack entry

[Doug Hardie]

pstack on radiusd (0.93) FreeBSD 4.6 shows what appears to be all 
threads idle except for this one:

----------------- thread 4 -----------------
 0x2820dcd8 _thread_kern_sched (0, 74756f, 73736553, 2820e652, 
28254da4, 28252e24) + 4c
 0x2820e6ab _thread_kern_sched_state_unlock (3, 8163f9c, 2824d100, fd, 
28254da4, 80c1d54) + 67
 0x28212d8c pthread_cond_wait (817a128, 817a124, bfac8b3c, 281d0b20, 
3bfc, eff) + 180
 0x281d0b6d sem_wait (80c1d54, bfac8bb0, bfac8b6c, 805ca79, bfac8bb0, 
aeff) + 69
 0x805cb03 rad_waitpid (aeff, bfac8bac, 0, 0, bfacb414, bfacbe3c) + 9b
 0x8056f89 radius_exec_program (817a0e0, 8173a00, 1, bfacbe3c, ff, 
816a400) + 66d
 0x8057f53 rad_authenticate (8173a00, 816a400, 8173a1c, 281d0b90, 
8163cd0, 8163cc0) + 89f
 0x8052c04 rad_respond (8173a00, 80576b4, 0, 0, 28254da4, bfabc000) + 
198
 0x805c21c _init (8163cc0, 0, 0, 0, 0, 0) + bb88
 0x281d24a3 _thread_start () + 37

That thread has said the same thing for over an hour.  It appears to be 
waiting for an Exec-Program-Wait module with pid of aeff to return.  
That pid is in the range where current processes were about an hour ago 
so it seems to match.  However, that process is long gone.  Somehow the 
signal was not caught.  Am I interpreting this correctly?  Every couple 
days radiusd just quits or hangs using all the processor.  When it 
hangs there are hundreds of threads.  I haven't check them carefully 
yet since this is a production system and having radiusd down causes 
all kinds of complaints.  I suspect that somehow this problem is 
growing slowly until eventually radiusd (or thread system) just can't 
handle the number of threads anymore.  Anyway thats a guess until I can 
capture the real thread state when it hangs.

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