Re: amvault hanging

On 05/19/2017 06:19 AM, Jean-Louis Martineau wrote:
> Orion,
> 
> Use attachment to attach the debug files. cut&paste are too much hard to read
> because the lines wrap.
> Can you get a stracktrace of all thread using gdb?
>   thread apply all bt
> 
> Jean-Louis

Another debug log with debug-recovery 9

Also when xfer-source-recovery.c:start_part_impl() is called it looks like
elt->offet == 0 and elt->orig_size == 0 so it calls:

    if (elt->offset == 0 && elt->orig_size == 0) {
        self->done = TRUE;
        g_mutex_unlock(self->start_part_mutex);
        return;
    }

(gdb) print *elt
$3 = {__parent__ = {g_type_instance = {g_class = 0x204f110}, ref_count = 2,
qdata = 0x0},
  xfer = 0x202c600, input_mech = XFER_MECH_NONE, output_mech =
XFER_MECH_PULL_BUFFER,
  upstream = 0x0, downstream = 0x205b030, cancelled = 0, expect_eof = 0,
can_generate_eof = 0,
  _input_fd = -1, _output_fd = -1, input_listen_addrs = 0x0,
output_listen_addrs = 0x0,
  repr = 0x203b6f0 "<XferSourceRecovery@0x2056000>", offset = 0, orig_size =
0, size = 0,
  block_size = 0, shm_ring = 0x0, crc = {crc = 4294967295, size = 0},
must_drain = 0,
  drain_mode = 0, cancel_on_success = 0, ignore_broken_pipe = 0}
(gdb) n
647             self->done = TRUE;

which doesn't seem quite right though I really don't know this code.

-- 
Orion Poplawski
Technical Manager                          720-772-5637
NWRA, Boulder/CoRA Office             FAX: 303-415-9702
3380 Mitchell Lane                       [email protected]
Boulder, CO 80301                   http://www.nwra.com