Red Hat Bugzilla – Bug 32122
Wolverine won't install on a 20GB disk
Last modified: 2008-05-01 11:38:00 EDT
I have a 20GB disk with 2490 cylinders. Disk Druid will not use this disk
during an install. RH 6.2 graphical install will allow you to use fdisk to
create a partition and Druid allows you ypu to select this one. When I
tried the Wolverine FTP install, it treated the fdisk warnings as fatal
and would not present the 20GB disk in the select partition dialog. Thus
I am unable to install.
What were the fdisk warnings?
[root@opal print]# fdisk /dev/hdb
The number of cylinders for this disk is set to 2490.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., LILO)
2) booting and partitioning software from other OSs
(e.g., DOS FDISK, OS/2 FDISK)
I always use cfdisk to create partitions and that gives no warnings. The XOSL
loader and latest lilo have no problems with large disks either. The FTP
install sees the warning (I guess) and presents a message box inviting you to
skip the disk!
Sorry, should have added fdisk does not like the partitions boundaries, this is
the error message install barfs with. Using the RH6.2 CDROM I get the same
messages but can proceed after creating the new partition (/dev/hdb4). I have
never experienced any problems with these partitions! The FTP install should
treat this situation as non-fatal.
[root@opal /root]# fdisk -l /dev/hdb
Disk /dev/hdb: 255 heads, 63 sectors, 2490 cylinders
Units = cylinders of 16065 * 512 bytes
Device Boot Start End Blocks Id System
/dev/hdb1 1 763 6126592+ 5 Extended
Partition 1 does not end on cylinder boundary:
phys=(762, 185, 63) should be (762, 254, 63)
/dev/hdb2 * 763 959 1572984 82 Linux swap
Partition 2 does not end on cylinder boundary:
phys=(958, 141, 63) should be (958, 254, 63)
/dev/hdb3 959 1089 1047784+ e Win95 FAT16 (LBA)
/dev/hdb4 1090 2090 8040532+ 83 Linux
/dev/hdb5 1 1 1953 78 Unknown
/dev/hdb6 1 16 124960+ 82 Linux swap
/dev/hdb7 16 265 1999840+ 83 Linux
/dev/hdb8 265 763 3999712+ 83 Linux
Could you append the output of /proc/ide/hdb/geometry when booted until 6.2?
Also from the 7.0 installer. To get this you need to switch to VC2 (cntl-alt-f2)
and then you can see a command prompt, where you can type
For some reason the 2.4 kernel seems to think your disk geometry is different
than the 2.2 kernel in 6.2 did.
I tried boot.img and bootnet.ing from Wolverine and neither allow Alt-F2 to get
to a command prompt. Also 'linux rescue' still puts you into the install.
On 6.2 the geometry is:
On 7.1 I can see the logical geometry is the same from the mount messages. I
think the difference is that with RH6.2 I used the graphical install which
allows you to use fdisk and then Druid. The geometry warnings are ignored. The
text install treats them as fatal.
Try these update images (make an ext2 floppy, copy these onto it, then
boot the installer like 'linux updates expert', and insert the floppy when
When it asks about a driver disk say 'No'.
Please let me know if this helps!
The files you want are:
Tried the update but got 'cpio failed on anaconda/newtpyfsedit.so: no space
left on device'. Looks like the RAM drive is full. Maybe you could create a
Could you try running 'strip' on the two .so files, should make them alot
smaller. I forgot to do that.
Try that out first.
I did the strip and now get 'error 2 reading header: cpio: bad magic'
I tried two diskettes and the same result. Any idea? I will be away until
Hmm we might be too different in versions for the update to work as I had hoped
I've tested this internally with a partition table like yours and it is working
fine. The final release will have this fix so you'll be able to install it.
Thanks for your help in resolving this issue.