Red Hat Bugzilla – Bug 33231
Installer fails to find suitable drives (on MegaRaid) to partition
Last modified: 2007-03-26 23:42:50 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)
When kickstarting (nfs), the process appears to proceed without any
problems until it is time to partition the drives. On this system I have 4
36gig scsi drives connected via adaptec 7890 Ultra 160/m controllers and
all the drives are under the control of a Dell MegaRaid Controller. At
parition time, the message "no suitable drives were found" (or something
very similiar, the word "suitable" is certainly there) is displayed and
the installation stops.
I have tried to poke about on the other virtual consoles, but I don't see
any helpful information.
This same machine can be kickstarted to RH7.0. I have also a "special"
RH7.0 kickstart image that contains the 2.4.2 kernel from rawhide - this
kickstarts okay too.
Steps to Reproduce:
1.Pretty much a vanilla kickstart. I include a "zerombr yes" and
a "clearpart --all" in the ks.cfg file
This same behaviour is found on two (very similiar) machines.
They are two and four processor Dell 6400's with the Dell PERC controller.
Boot message from a 7.0 kernel tell me these are AMI MegaRAID 1.01 254
commands). /proc/scsi/scsi tells me it is a MegaRAID Model: LD0 RAID5
0403R Rev: 1.01) These machines have from 1 to 2 gig ram, 4x36gig drives
(stiped into one big 100gig file system), two Adaptec 7899 Ultra 160/m
controllers in each machine.
When you get to the message window that says it can't find drives to install on,
switch to Alt+F2 and run "cat /proc/scsi/scsi" and "lsmod" to make sure that
megaraid was loaded.
I am sorry I overlooked adding that detail. Indeed, the driver is NOT loaded.
on the same console (Alt+F2), please send me the output of "lspci; lspci -n"
Created attachment 13705 [details]
lspci ; lspci -n output
That lspci output attachment was just created on one of the the systems that
has this trouble but is currently running rh7.0, kernel 2.2.16-22smp
I will boot it with kickstart and verify the details are the same.
Created attachment 13708 [details]
Created attachment 13709 [details]
I verified the output of lspci and lspci -n are the same when the system is in
anaconda during kickstart as when running rh7.0
can you add "lspci -vv" output to this report as well?
Bill - can you verify that the correct module would load in this system?
Yes, it's in the pcitable. Try switching the megaraid out of i2o mode.
Created attachment 13827 [details]
lspci -vv output from two machines
it seems you have your megaraid in I2O mode. Go into the BIOS and turn that
I2O did not have any effect. The other option was "Mass Storage". After changing
this parameter, the install still failed.
While preparing new lspci dumps for you I thought to cat /proc/version. This
showed me I was running a 2.2.16-22BOOT kernel dated from Aug22/2000. This
suggested to me that my problem might be with the kickstart floppy itself. I
have been using this same floppy for quite a while. Why this same floppy would
work okay for RH7.0 and Rawhide, I cannot imagine. But it does.
So, I created a new boot floppy from the wolverine bootnet.img file.
So to summarize, it looks like there is some dependancy between the bootnet.img
used to kickstart the system and the version of Redhat that is to be installed,
at least this seems so with wolverine.
Your boot image should always match the tree you are installing from.
Sorry for the confusion.