Skip to contentSkip to conversation

md: bug in file drivers/md/md.c, line 1513

DISCUSSION Paul Slootman 2 participants 3 messages Mar 22, 2005 – Mar 24, 2005

lore
Share & tools
lore
#0 lore raw

This is on kernel 2.6.11, mdadm 1.4.0

The system has MD devices that are auto-configured on boot.

However, there are also devices connected via another SCSI adapter (actually, a Qlogic QLA2300). I'm using a module for that. As the auto-configure only runs at boot (or rather, when the md subsystem is started). I wanted to restart a raid-0 device that I had previously created. I did:

	mdadm --run /dev/md10

as a simple attempt to see what would happen. What happened was the error message in the subject, and a "COMPLETE RAID STATE PRINTOUT"... In that output there is a line "md10:", the next line is "md1: <sde1><sdd1><sdc1><sdb1><sda1>".

Admittedly the usage may be wrong, but having the kernel say "bug" can't be right :-)

Paul Slootman

#1 lore raw
Quoted 11 lines · On Tuesday March 22, [email protected] wrote:

This is on kernel 2.6.11, mdadm 1.4.0

The system has MD devices that are auto-configured on boot.

However, there are also devices connected via another SCSI adapter (actually, a Qlogic QLA2300). I'm using a module for that. As the auto-configure only runs at boot (or rather, when the md subsystem is started). I wanted to restart a raid-0 device that I had previously created. I did:

	mdadm --run /dev/md10

As you admit, this is wrong. You want something like
mdadm --assemble /dev/md10 /dev/.....(list of component devices)

or describe the md10 array (e.g. via UUID) in /etc/mdadm.conf

Quoted 10 lines

as a simple attempt to see what would happen. What happened was the error message in the subject, and a "COMPLETE RAID STATE PRINTOUT"... In that output there is a line "md10:", the next line is "md1: <sde1><sdd1><sdc1><sdb1><sda1>".

Admittedly the usage may be wrong, but having the kernel say "bug" can't be right :-)

Yes, there are quite a few of those silly bug messages. I've removed a few, but have not yet gone through and checked and removed all the bad ones.
NeilBrown

#2 lore raw
Quoted 11 lines · Neil Brown
Quoted 10 lines · On Tuesday March 22, [email protected] wrote:

This is on kernel 2.6.11, mdadm 1.4.0

The system has MD devices that are auto-configured on boot.

However, there are also devices connected via another SCSI adapter (actually, a Qlogic QLA2300). I'm using a module for that. As the auto-configure only runs at boot (or rather, when the md subsystem is started). I wanted to restart a raid-0 device that I had previously created. I did:

Quoted 6 lines
Quoted 1 line
	mdadm --run /dev/md10

As you admit, this is wrong. You want something like
mdadm --assemble /dev/md10 /dev/.....(list of component devices)

or describe the md10 array (e.g. via UUID) in /etc/mdadm.conf

It took a bit of trial and error (I find the mdadm docs a bit confusing...) but I came up with this:

mdadm --assemble --super-minor=10 --config=partitions /dev/md10

Quoted 5 lines
Quoted 5 lines

error message in the subject, and a "COMPLETE RAID STATE PRINTOUT"... In that output there is a line "md10:", the next line is "md1: <sde1><sdd1><sdc1><sdb1><sda1>".

Quoted 7 lines
Quoted 3 lines

Admittedly the usage may be wrong, but having the kernel say "bug" can't be right :-)

Yes, there are quite a few of those silly bug messages. I've removed a few, but have not yet gone through and checked and removed all the bad ones.

Keep up the good work :-)

Paul Slootman