Skip to contentSkip to conversation

device failure hangs the system

DISCUSSION Anand Jain 3 participants 5 messages Jul 1, 2011 – Jul 7, 2011 Btrfs file system

lore
Share & tools
lore

Anand Jain

author
#0 lore raw

hi,

Following test case causes my remote system to hard-hang and does not respond to any key strokes.

-----------

# btrfs fi show
failed to read /dev/sr0
Label: none  uuid: 75ad3c9f-f661-498e-8c13-89d4e4c58312
	Total devices 3 FS bytes used 28.00KB
	devid    1 size 465.76GB used 2.02GB path /dev/sdb
	devid    3 size 465.76GB used 2.01GB path /dev/sdd
	devid    2 size 465.76GB used 1.01GB path /dev/sdc

Btrfs v0.19-35-g1b444cd
------------

Stopping the disk
------------

# echo 1 > /sys/block/sdd/device/delete
------------
------------
# dmesg | tail
sd 3:0:0:0: [sdd] Stopping disk
::
#
------------

------------

# btrfs fi show
failed to read /dev/sr0
Label: none  uuid: 75ad3c9f-f661-498e-8c13-89d4e4c58312
	Total devices 3 FS bytes used 28.00KB
	devid    1 size 465.76GB used 2.02GB path /dev/sdb
	devid    2 size 465.76GB used 1.01GB path /dev/sdc
	*** Some devices missing

Btrfs v0.19-35-g1b444cd
------------

and the following command hangs the system.
-------------

# btrfs fi balance /btrfs
-------------

Any idea if this is an known issue ? OR if there is any better way to fail a disk (or a loop disk) for testing.

(something similar to cfgadm -c unconfigure in solaris).

Thanks
-Anand

author
#1 lore raw

Looks like there is a panic (not system hang).
any idea where is the panic log after the system has been power-recycled. (its not in the /var/log/messages or dmesg or /var/crash is empty)

Thanks, Anand

Quoted 55 lines · Anand Jain

hi,

Following test case causes my remote system to hard-hang and does not respond to any key strokes.

-----------

# btrfs fi show
failed to read /dev/sr0
Label: none uuid: 75ad3c9f-f661-498e-8c13-89d4e4c58312
Total devices 3 FS bytes used 28.00KB
devid 1 size 465.76GB used 2.02GB path /dev/sdb
devid 3 size 465.76GB used 2.01GB path /dev/sdd
devid 2 size 465.76GB used 1.01GB path /dev/sdc

Btrfs v0.19-35-g1b444cd
------------

Stopping the disk
------------

# echo 1 > /sys/block/sdd/device/delete
------------
------------
# dmesg | tail
sd 3:0:0:0: [sdd] Stopping disk
::
#
------------

------------

# btrfs fi show
failed to read /dev/sr0
Label: none uuid: 75ad3c9f-f661-498e-8c13-89d4e4c58312
Total devices 3 FS bytes used 28.00KB
devid 1 size 465.76GB used 2.02GB path /dev/sdb
devid 2 size 465.76GB used 1.01GB path /dev/sdc
*** Some devices missing

Btrfs v0.19-35-g1b444cd
------------

and the following command hangs the system.
-------------

# btrfs fi balance /btrfs
-------------

Any idea if this is an known issue ? OR if there is any better way to fail a disk (or a loop disk) for testing.

(something similar to cfgadm -c unconfigure in solaris).

Thanks
-Anand

#2 lore raw
Quoted 4 lines · On Fri, 01 Jul 2011 17:42:33 +0800 Anand Jain <[email protected]> wrote:

Looks like there is a panic (not system hang).
any idea where is the panic log after the system has been power-recycled. (its not in the /var/log/messages or dmesg or /var/crash is empty)

You can get it by seting up netconsole logging to another machine and then reproducing the crash.
http://www.cyberciti.biz/tips/linux-netconsole-log-management-tutorial.html

Signature · 2 lines
With respect,
Roman
#3 lore raw
In reply to Anand Jain
Quoted 15 lines · Anand Jain

hi,

Following test case causes my remote system to hard-hang and does not respond to any key strokes.

-----------

# btrfs fi show
failed to read /dev/sr0
Label: none  uuid: 75ad3c9f-f661-498e-8c13-89d4e4c58312
    Total devices 3 FS bytes used 28.00KB
    devid    1 size 465.76GB used 2.02GB path /dev/sdb
    devid    3 size 465.76GB used 2.01GB path /dev/sdd
    devid    2 size 465.76GB used 1.01GB path /dev/sdc
Quoted 8 lines

Btrfs v0.19-35-g1b444cd
------------

Stopping the disk
------------

# echo 1 > /sys/block/sdd/device/delete
------------

Well that's a neat trick, do you have a way to undo that action too? Seems a rescan doesn't make it show back up. Please try the patch I just posted to the list to fix this problem. Thanks,

Josef

author
#4 lore raw

Josef,

Quoted 2 lines

Well that's a neat trick, do you have a way to undo that action too? Seems a rescan doesn't make it show back up.

hope the following helps..
-------------

 # fdisk -l /dev/sdg | egrep "Disk /"
 Disk /dev/sdg: 4294 MB, 4294967296 bytes

# x=ls -l /sys/class/block/sdg | cut -d "/" -f12 | sed 's/:/ /g'

 # echo "scsi remove-single-device ${x}" > /proc/scsi/scsi
 # fdisk -l /dev/sdg | egrep "Disk /"
 # echo "scsi add-single-device ${x}" > /proc/scsi/scsi
 # fdisk -l /dev/sdg | egrep "Disk /"
 Disk /dev/sdg: 4294 MB, 4294967296 bytes
-------------
Quoted 1 line

Please try the patch I just posted to the list to fix this problem. Thanks,

Facing some challenges to upgrade my machine to 3.0.0-rc6, so is the delay.

Thanks for the patch.

Anand

Quoted 1 line

Josef