Bacula-users

Re: [Bacula-users] Shutting down SD / blocks kernel

2011-01-25 00:55:49
Subject: Re: [Bacula-users] Shutting down SD / blocks kernel
From: Jesper Krogh <jesper AT krogh DOT cc>
To: bacula-users AT lists.sourceforge DOT net
Date: Tue, 25 Jan 2011 06:52:04 +0100
Jan 20 15:40:34 kernel: INFO: task bacula-sd:28017 blocked for more
> than 120 seconds.
> Jan 20 15:40:34 kernel: "echo 0>  /proc/sys/kernel/
> hung_task_timeout_secs" disables this message.
> Jan 20 15:40:34 kernel: bacula-sd     D 000d6fd6dc61fc69     0
> 28017      1               30457 (NOTLB)
> Jan 20 15:40:34 kernel:  ffff8800a2ccbcd8  0000000000000282
> ffff8800f8de2000  ffffffff8807b185
> Jan 20 15:40:34 kernel:  000000000000000a  ffff8800d7051860
> ffff880009f64080  0000000000039dbb
> Jan 20 15:40:34 kernel:  ffff8800d7051a48  ffffffff80336433
> Jan 20 15:40:34 kernel: Call Trace:
> Jan 20 15:40:34 kernel:  [<ffffffff8807b185>] :scsi_mod:scsi_request_fn
> +0x347/0x39c
> Jan 20 15:40:34 kernel:  [<ffffffff80336433>] blk_execute_rq_nowait
> +0x89/0xa0
> Jan 20 15:40:34 kernel:  [<ffffffff8022d660>] wake_up_bit+0x11/0x22
> Jan 20 15:40:34 kernel:  [<ffffffff80262fb3>] wait_for_completion+0x7d/
> 0xaa
> Jan 20 15:40:34 kernel:  [<ffffffff80288e5e>] default_wake_function
> +0x0/0xe
> Jan 20 15:40:34 kernel:  [<ffffffff881dbe23>] :st:st_do_scsi+0x1f4/0x221
> Jan 20 15:40:34 kernel:  [<ffffffff881dc932>] :st:st_int_ioctl
> +0x5f2/0x92b
> Jan 20 15:40:34 kernel:  [<ffffffff881dea58>] :st:st_ioctl+0xaa5/0xe1f
> Jan 20 15:40:34 kernel:  [<ffffffff8023a4a7>] may_delete+0x69/0x138
> Jan 20 15:40:34 kernel:  [<ffffffff80243fc9>] do_ioctl+0x55/0x6b
> Jan 20 15:40:34 kernel:  [<ffffffff802316c1>] vfs_ioctl+0x457/0x4b9
> Jan 20 15:40:34 kernel:  [<ffffffff802afcf3>] audit_syscall_entry
> +0x180/0x1b3
> Jan 20 15:40:34 kernel:  [<ffffffff8024e52b>] sys_ioctl+0x59/0x78
> Jan 20 15:40:34 kernel:  [<ffffffff802602f9>] tracesys+0xab/0xb6
>
>
> it seems that the tape drive (LTO-3 TANBERG/DELL) needs more time that
> the sd-daemons expects !?

I'be been told, by a tape library technichian, than LTO incorporates
a small chip on the tapes, that will have an index of what and where
stuff is on the tape. If the process gets unclean shut down (power-failure
sd-crash or similar) then then it will know that the chip isn't current and
have to scan the entire tape to update the chip before any actions
can take place. This looks like the daemon is haning to the OS as
is in fact taking way longer than 120s.

That perfectly fits the cases where I see the above messages on
my storage daemon.

-- 
Jesper

------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires 
February 28th, so secure your free ArcSight Logger TODAY! 
http://p.sf.net/sfu/arcsight-sfd2d
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

<Prev in Thread] Current Thread [Next in Thread>