[Bacula-bugs] [bacula 0000243]: SD crashes when running multiple concurrent backups

[<bac...@li...>]

The following bug has been CLOSED
======================================================================
http://bugs.bacula.org/bug_view_advanced_page.php?bug_id=0000243
======================================================================
Reported By:                ifmpeter
Assigned To:                
======================================================================
Project:                    bacula
Bug ID:                     243
Category:                   Storage Daemon
Reproducibility:            always
Severity:                   crash
Priority:                   normal
Status:                     closed
======================================================================
Date Submitted:             02-23-2005 03:50 PST
Last Modified:              03-02-2005 12:54 PST
======================================================================
Summary:                    SD crashes when running multiple concurrent backups
Description: 
I had a full backup running and then initiated another backup
on another client, in another pool, using another tape drive
then bacula-sd coredumped at:


  [1] 0xfefbc210(0x22d3, 0x0, 0x0, 0x22d3, 0xfee80c00, 0x0), at 0xfefbc20f

  [2] 0xfefafab0(0x22d3, 0x0, 0x0, 0x0, 0xffbffeff, 0xfee80c00), at
0xfefafaaf 
  [3] signal_handler(sig = ???) (optimized), at 0x4b430 (line ~159) in
"signal.c"
  [4] 0xfefbb04c(0xb, 0x0, 0xfeafba58, 0x4b0bc, 0x0, 0x1), at 0xfefbb04b 
  [5] 0xfefb08e4(0xb, 0xffbffeff, 0x0, 0x0, 0xfee80c00, 0xfeafba58), at
0xfefb08e3 
=>[6] reserve_device_for_append(jcr = ???, dev = ???) (optimized), at
0x1e4b4 (line ~372) in "acquire.c"
  [7] use_device_cmd(jcr = ???) (optimized), at 0x2d2e0 (line ~308) in
"job.c"
  [8] use_cmd(jcr = ???) (optimized), at 0x2cbf0 (line ~145) in "job.c"
  [9] handle_connection_request(arg = ???) (optimized), at 0x28b58 (line
~204) in "dircmd.c"
  [10] workq_server(arg = ???) (optimized), at 0x4e9e8 (line ~347) in
"workq.c"


  366          }
  367      } else {
  368         if (dev->can_read()) {
  369            Jmsg(jcr, M_FATAL, 0, _("Device %s is busy reading.\n"),
dev_name(dev));
  370            goto get_out;
  371         }
  372         ASSERT(dev->num_writers == 0);
  373      }
  374   
  375      dev->reserved_device++;

======================================================================

----------------------------------------------------------------------
 kern - 02-25-2005 03:55 PST 
----------------------------------------------------------------------
Please try the same thing with the current CVS. I believe that I have
corrected the problem. However, there are still a number of issues
relating to "reserving" drives that I have not yet had the time to work
out. I should be making some significant improvements in this over the
next week.

----------------------------------------------------------------------
 ifmpeter - 02-25-2005 14:14 PST 
----------------------------------------------------------------------
It looks better now I think - atleast it hasn't coredumped on me yet... :)

----------------------------------------------------------------------
 kern - 02-26-2005 14:11 PST 
----------------------------------------------------------------------
I've corrected a few more problems in the most recent CVS as well.  

There are still problems where a drive is not available and the job will
fail. Hopefully, some changes I am making now will make the job wait until
a drive is available.

----------------------------------------------------------------------
 kern - 03-02-2005 12:54 PST 
----------------------------------------------------------------------
I'm closing this because the problem is most likely fixed.  If you have any
more problems please re-open this bug or a new one ...

Bug History
Date Modified  Username       Field                    Change              
======================================================================
02-23-05 03:50 ifmpeter       New Bug                                      
02-25-05 03:55 kern           Bugnote Added: 0000694                       
02-25-05 03:55 kern           Status                   new => feedback     
02-25-05 03:55 kern           version                  1.37.2 => 1.37.4    
02-25-05 14:14 ifmpeter       Bugnote Added: 0000698                       
02-26-05 14:11 kern           Bugnote Added: 0000706                       
03-02-05 12:54 kern           Bugnote Added: 0000713                       
03-02-05 12:54 kern           Resolution               open => fixed       
03-02-05 12:54 kern           Status                   feedback => closed  
======================================================================