You should reboot the computer.

The reason I suggested using cfdisk program because a user cannot create an extended partition in cfdisk. Whenever the first logical partition is created then the extended partition will be formed as a by-product. An extended partition has no storage unit inide but just acts as an envelope for all the logical partitions inside.

Basically a hard disk can have a mximum of 4 primary partition. Each has the the hard disk address of the begining and fisnishing point of the partition. If a user want more than 4 partition then one of the 4 primaries must be given up to turn into an extended partition. This is done by altering the partition type but the begining and finishing points are defined exactly like a primary. Inside the exetnded partition one logical partition must carries the address for the next one up and all of them must be in a continuous chain, with the end link matching the finishing point specified in the extended partition. A logical partition at the middle if deleted will have the space "dead" and Linux will automatically shift everything up. This phenomanon can be seen in cfdisk program.

It is therefore not possible to format an extended partition and usch a misadventure could not have happened if cfdisk program was used.

Michaelk's recommendation is a good practice that any alteration to the partition table can only be secured after a reboot so that the operating system can use the new confiuration.

I cannnot find "cfdisk" command under my Red Hat Linux...where is it?

Data Sheet

Ah yes the Red Hat family is the odd one out that uses sfdisk instead of cfdisk. It isn't the same thing. Try cfdisk on a Live CD.

I think both fdisk and sfdisk permit a user to declare an extended partition and assume the user knows to to use it.

The major advantage of cfdisk is you can see the partition names and the relative positions in the hard disk in a tabulated format. This is important to a user who creates a logical partition first and then a primary second. This is because the first logical partition must be hda5 (or sda5 in a Sata) and the first available primary partition hda1 would be converted to the extended partition necessary if any logical partition is formed, without the user intervention. The actual first primary partition created must take the name hda2 and follow immediately behind hda5 in the hard disk.

The above may sound crazy but that is exactly the partitions would be arranged in the PC hard disk if a user wants one logical partition first and then a primary partition second.

While the partition table has been updated, the kernel still works on the old one. Thus the refusal to create a new filesystem. When you re-boot, it should all work.

I use "raw device" to mean anything accessible thru a low-level utility such as dd---without the benefit of any filesystem structure. Whether or not a disk has been partitioned or had a filessytem installed it always accessible as a raw device.