I fdisk'ed hdc to divide a 15G partition into two, so as to have somewhere for SuSE 8.2 pro and Gentoo, and fdisk wrote with a

"The partition table has been altered!

Calling ioctl() to re-read partition table.

WARNING: Re-reading the partition table failed with error 16: Device or resource busy.
The kernel still uses the old table.
The new table will be used at the next reboot.
Syncing disks."

So. ( "cat /proc/partitions" confirmed this )

Then I tried "apropos ioctl", and got to doing this:

"sync
blockdev --flushbufs /dev/hdc
blockdev -v --rereadpt /dev/hdc"

repeatedly, and the kernel /still/ wouldn't read the new partition table.

My question is HOW does one force the 2.4.x kernel to reread the partition-table?

( with the implicit
"being forced to reboot whenever changing the partition-table means non-mission-critical, as far as I can see..."
... or are we forced to use LVM instead of regular partitioning? In spite of the kernel-hackers' low opinion of the current version of LVM? )

TIA..

maybe you can do it by booting into single user mode and then unmounting the partition you have altered

that could work, im not sure though...

the partition I altered wasn't mounted ( I'm not THAT stupid, usually
: D )

I do have a RAID5 array using the drive, though

( other partitions:
primary-3 and primary-4 of each drive participate in a raid5 array, all primary-3's in one, all primary-4's in another .. this way I don't have to worry about re-writing /etc/raidtab every time I re-write the partition-table to modify the logicals in the extended parition2 of any drive, see )

-sigh- it seems the arrays're going to have to be unmounted to get the kernel to be able to re-load the change in the extended-partition table... grr