[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Problem with external RAID



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jesus Roncero wrote:
> Hi, I've been fighting a couple of days trying to set up a debian
> machine and an external RAID to work with debian. The
> configuration is: A dell poweredge 850 server with two serial ata
> drives. A LSI PCI SCSI Controller, that works under linux. a
> EonStor RAID, connected to the scsi controller using a cable.
> 
> the thing is, after using the last kernel, we managed to make the
> scsi card work correctly. Apparently there are some bugs in the
> previous versions of the kernel/drivers for the scsi card that
> create some problems when trying to access a device. We had those
> with an earlier version of the kernel and got it fixed by using
> 2.6.17.4.
> 
> So, we are using etch on this server in order to recognize all
> the hardware. Everything is working flawlessly (apparently). The
> sata drives are mapped by default to /dev/sda and /dev/sdb and
> that's how we installed debian. But the problem arises when we
> connect the external RAID to the system. The thing is that the
> system randomly assigns the device /dev/sdc and /dev/sda to the
> RAID, so, sometimes it boots perfectly (when it is /dev/sdc) and
> some others, the boot process aborts (when it is /dev/sda)
> because the linux is not finding the / partitions (as the sata
> disks are now located at /dev/sdb and /dev/sdc).
> 
> From what I have been reading, the system loads the modules
> asynchronously using udev (I guess) and so that's why the RAID is
> assigned /dev/sda or /dev/sdc. I've been trying to figure out how
> to set the order at which I'd like the modules to be loaded, but
> with no luck. I've been trying to define a couple of rules at
> /etc/udev/rules.d/ in order to change the devices assigned, but
> all I get are some symlinks to the real devices in /dev but the 
> problem continues to be there.

That's typical, I think, of Linux and SCSI.

Are the drivers modules or compiled-in?

What if you set the RAID device to be noauto mount, and then create
a script in /etc/rcS.d somewhere after /etc/fstab is processed.
That script would "manually" mount the RAID device.

> So, I would really appreciate if someone could give some hints on
> how to alter the order at which the devices are defined as I am a
> little bit lost. I don't know if it has something to do with
> initrd images, or udev, hotplug or the order at which the modules
> are loaded at boot time. Any hint would be appreciated, ;)
> really.

Also maybe you could write udev rules to avoid the issue by giving
symbolic names to all the drives.

> I've also been thinking about getting rid of udev, but I don't
> want to create a complete mess as I don't really know if this is
> really I should be doing.



- --
Ron Johnson, Jr.
Jefferson LA  USA

Is "common sense" really valid?
For example, it is "common sense" to white-power racists that
whites are superior to blacks, and that those with brown skins
are mud people.
However, that "common sense" is obviously wrong.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEruDkS9HxQb37XmcRAt8SAKCb8NMlZusaWw+BKBt3MzIJO5mDGwCggmIK
UpOE0faf84wiJVUc2EjZ5qE=
=6aNj
-----END PGP SIGNATURE-----



Reply to: