A follow-up. After a reboot, the wrongly sized raid failed to autorun.
A mkraid on the same partitions got the raid5 size up to the correct size.
What I think happened was that I had re-fdisked the raid partition to be
the new size, ignored the (you should reboot to make sure the new
partition has been correctly updated on disk) message, and even though
fdisk reported the correct size for the new partitions, mkraid used the
old size for the partition (even though I had zeroed it).
On Thu, 4 Feb 1999 [EMAIL PROTECTED] wrote:
>
> Using Linux 2.2.1 and the raidtools .90 patch there appears to be an
> overflow bug in dealing with >2 gb disk partitions on the i386
> architecture.
>
> I have 4 drives laid out more or less as follows:
>
> /dev/sda1 * 1 5 21824 83 Linux native
> /dev/sda2 6 1016 4419081 5 Extended
> /dev/sda5 6 99 410843 83 Linux native
> /dev/sda6 100 127 122357 82 Linux swap
> /dev/sda7 128 409 1232591 83 Linux native
> /dev/sda8 410 1016 2653166 fd Unknown
>
> When I raid5 together /dev/sd*8 I get a total raid array size of:
>
> Filesystem 1024-blocks Used Available Capacity Mounted on
> /dev/md0 1498416 52 1421020 0% /mnt/raid
>
> which is, um, quite a bit smaller than I expected. Gee, there
> must be quite a bit of ecc code in there or something... :)
>
> If I'm doing something else stupid (old e2fs tools, mount,
> etc - this is a plain redhat 5.2 system with the new modutils and
> 2.2.1 kernel only) please let me know.
>
> ________
> mike
>
>
