FC2 & FC3 SCSI issues

[Thomas J. Baker]

On Thu, 2004-10-21 at 14:32 -0400, Bill Cronk wrote:
> I have not resolved my SCSI RAID box problems when running either FC2 or 
> FC3t2 or FC3t3. I have been reading over past messages in the 
> fedora-test-list archives and have not found anything related yet to the 
> problem I am experiencing and am having difficulty resolving.
> 
> So let me reload my issues here to the list:
> 
> First off I have multiple RAID boxes originally setup on seperate 
> servers. I now have new servers which I want to put into my network, but 
> due to the problems getting the RAID boxes to load correctly I can't.
> 
> The problem(s): Originally I had the hard drive ID across all drives in 
> a RAID set to the default linux setting, did not have them configured 
> with 'fd' ID. These RAID boxes work on older installations and two of 
> them work on one server which has the full FC2 install.... I don't know 
> why they work and I can not duplicate the setup on new installs. They do 
> have different motherboards between the two servers.
> 
> On the latest test server I have FC3 installed, built one new RAID box 
> of 14 drives in a JBOD configuration with software raid5, and added on 
> the second SCSI port another RAID box setup as raid0 from another 
> server. I can make both start up and work by running a script which does 
> what should happen in the boot cycle. If I leave the normal raidtab file 
> in place then all fails and the boot cycle drops into single user mode 
> for repairs to the system. I believe now that this problem could have 
> been overcome by removing the spaces in the raidtab file at the 
> beginning of the 'raid-level' statement.
> 
> However, following a suggestion from a person here in the 
> fedora-test-list, I changed the drive ID to 'fd' and I started to try 
> using mdadm to mange the RAID which when combined now opens another 
> whole set of problems for me. One RAID box works fine on the Adaptec 
> U320 add on card, two RAID boxes starts the confusion and results in one 
> box not setting up due to the /dev/md1 device file missing. Tried 
> creating the /dev/md1 then setting up the RAID and it all works just 
> fine. Reboot the computer and /dev/md1 is gone.
> 
> So how is one supposed to fix this type of a problem which basically 
> revolves around the missing /dev/md1, but my impression is that between 
> udev and hotplug this should not have happened. Shouldn't the additional 
> SCSI RAID device being detected have automatically setup the device file 
> needed and connected to it? What needs to happen then if I needed to 
> plug onto three out of four SCSI ports on this box? Currently I believe 
> nothing would happen... it would just not work.
> 
> Any suggestions??
> 
> Sorry for being long winded...
> 
> Bill
> 
> 

Did you set up the mdadm.conf file and describe all the arrays in it?

It's been a while since I set mine up and after converting to the 'fd'
partition type, it always just works without needing to configure the
mdadm.conf file. I only have one array though. I would guess that if you
set up the mdadm.conf file and used mdadm --assemble, possibly in the
rc.local file, it might at least pick up the second array.

tjb
-- 
=======================================================================
| Thomas Baker                                  email: tjb at unh.edu    |
| Systems Programmer                                                  |
| Research Computing Center                     voice: (603) 862-4490 |
| University of New Hampshire                     fax: (603) 862-1761 |
| 332 Morse Hall                                                      |
| Durham, NH 03824 USA              http://wintermute.sr.unh.edu/~tjb |
=======================================================================