Bind 9.4.3-P3 on Solaris 10 Hang

Steve Foster s.p.foster at leeds.ac.uk
Mon Dec 7 11:45:43 UTC 2009


On Fri, 4 Dec 2009, Steve Foster wrote:

Hi Folks,

just to followup on this, I've caputred a core of named when this happens. 
>From the looks of it there is a deadlock/very long pause occuring as most 
of the threads appear to be parked.

The output of the stacktrace suggests that 3 out of 5 threads are parked 
for soem reason.

Anyone got any suggestions?

Cheers

Steve

Here is a stack trace of the threads:

stack pointer for thread 1: ffffffff7fffee51
[ ffffffff7fffee51 libc.so.1`___sigtimedwait+4() ]
   ffffffff7ffff011 libc.so.1`__posix_sigwait+0x18()
   ffffffff7ffff0c1 isc_app_run+0x268()
   ffffffff7ffff221 main+0x208()
   ffffffff7ffff2e1 _start+0x17c()
stack pointer for thread 2: ffffffff7db0f0e1
[ ffffffff7db0f0e1 libc.so.1`__lwp_park+0x10() ]
   ffffffff7db0f191 dns_dispatch_attach+0xbc()
   ffffffff7db0f241 fctx_query+0x3c4()
   ffffffff7db0f431 fctx_try+0x19c()
   ffffffff7db0f4f1 fctx_timeout+0x174()
   ffffffff7db0f5d1 dispatch+0x48c()
   ffffffff7db0f6a1 run+4()
   ffffffff7db0f751 libc.so.1`_lwp_start()
stack pointer for thread 3: ffffffff7da0e7e1
[ ffffffff7da0e7e1 libc.so.1`_write+8() ]
   ffffffff7da0e891 select_poke+0x28()
   ffffffff7da0e9d1 socket_recv+0x190()
   ffffffff7da0eac1 startrecv+0x194()
   ffffffff7da0eb81 dns_dispatch_addresponse2+0x958()
   ffffffff7da0ec61 resquery_send+0x158()
   ffffffff7da0f241 fctx_query+0x518()
   ffffffff7da0f431 fctx_try+0x19c()
   ffffffff7da0f4f1 fctx_timeout+0x174()
   ffffffff7da0f5d1 dispatch+0x48c()
   ffffffff7da0f6a1 run+4()
   ffffffff7da0f751 libc.so.1`_lwp_start()
stack pointer for thread 4: ffffffff7d90f121
[ ffffffff7d90f121 libc.so.1`__lwp_park+0x10() ]
   ffffffff7d90f1d1 libc.so.1`cond_wait_queue+0x28()
   ffffffff7d90f281 libc.so.1`cond_wait_common+0x2d8()
   ffffffff7d90f331 libc.so.1`_cond_timedwait+0x34()
   ffffffff7d90f3f1 libc.so.1`cond_timedwait+0x14()
   ffffffff7d90f4a1 libc.so.1`pthread_cond_timedwait+0xc()
   ffffffff7d90f551 isc_condition_waituntil+0x9c()
   ffffffff7d90f691 run+0xc0()
   ffffffff7d90f751 libc.so.1`_lwp_start()
stack pointer for thread 5: ffffffff7d80f3f1
[ ffffffff7d80f3f1 libc.so.1`__lwp_park+0x10() ]
   ffffffff7d80f4a1 process_fd+0x158()
   ffffffff7d80f551 process_fds+0x108()
   ffffffff7d80f601 watcher+0x138()
   ffffffff7d80f751 libc.so.1`_lwp_start()

So the only threads that are not parked are what looks to be the parent 
thread and a results thread.




More information about the bind-users mailing list