FC3t3 -- udev and SCSI devices

[ngc4013 at cox.net]

Harald wrote:

 

> what is printed in /var/log/messages for your second RAID box?

> In rc.sysinit:

          /sbin/mdadm –A –s

> is run if /etc/mdadm.conf is present


 

First there is an error in the dmesg output that I find to be intense:

 

            Program scsi_unique_id is using a deprecated scsi ioctl, please convert it to SG_IO

 

Many many lines of this message


 

 

As for my RAID boxes here is what I have found out so far:

 

If I run at the prompt ‘/sbin/mdadm –A –s’ it exits with the following message:

 

            mdadm: error opening /dev/md1: No such file or directoy

 

I decided to update the FC3t3 to FC3rc1 and discovered something interesting which sheds some light on the problem. When disk druid started up and I flagged the two partitions I needed to reformat and set the mount points, I noticed for the first time, since I currently had both RAID boxes connected yet, that the two RAID boxes showed up in the drive listing. Now I can see what part or possibly all of the problems I am seeing are caused by. Both RAID boxes are listed as /dev/md0
 obviously they both can not operate as /dev/md0, but now it makes sense.  I figure that when creating a RAID box part of the superblock structure carries the info regarding the drive configuration and /dev/md? tag.

A little history about the RAID boxes, both were created on separate servers as /dev/md0 connections.

I would assume that when the RAID box is auto detected that it reads in the md0 tag and passes it to udev which now can not setup a connection because the md0 tag is already accounted for with the first RAID box which also accounts for the lack of the /dev/md1 file being created.

This also means that if I move the other two RAID boxes over to the U160 SCSI ports that now the problem will be recreated again two more times since they are also setup as /dev/md0.

So now the question is how can the md0 tag be changed to an md1 tag without destroying the data on the RAID box and will it actually then fix the problem?

BTW, if you need them I can email the configuration files directly to you such as the /etc/mdadm.conf file.

Also, why does the system insist on automatically creating /media/scsidevice and /media/scsidevice1 mount points for the first SCSI drive from each RAID box?

Bill