[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: bugs in partman-auto_41 (trying to create 3 primary partitions with preseed)

On Tue, Feb 14, 2006 at 05:38:10PM +0100, Steven Van Acker wrote:
> I want to create 3 primary partitions using a preseeded
> recipe. partman-auto creates the first 2 partitions without problem, but
> errors on the 3rd. The reason is that it tries to create a partition
> that is bigger than the remaining free diskspace.
> Considering that a request only has to be 1 byte over the allowed
> size to make partman-auto complain, being numerically very correct is a
> prerequisite to calculating correct partition sizes.

If you want to specify the exact sizes of the partitions, then for the
first and the second partition use minsize=maxsize=the size you want.
And for the third partition use minsize=0 maxsize=10000000.

> - The first partition doesn't start from byte 0 but from byte 32256.
>   This is only about 32KB, but big enough to cause problems. As far as I
>   can see, partman-auto doesn't take that into consideration.

This is normal behaviour of parted in order to keep the partition
table compatible with DOS and to reserve enough free space for boot

> - Sizes expressed in bytes are rounded to megabytes by dropping the last
>   6 digits. Shouldn't they be divided by 1024*1024 ?

The manifacturers of the disks use decimal multipliers.  The binary
multipliers are useful mostly to measure the size of the RAM.

> - Asking for a partition of N bytes does not mean you get exactly N
>   bytes. In my case, I wanted a partition of 64MB for /boot.
>   partman-auto translated that to 64000001 bytes and when the partition
>   was finally created, it was 65802240 bytes. That's a lot of extra
>   bytes.

Whenever possible the partitions are aligned at the cylinder
boundaries (for compatibility with DOS).  If the disk uses LBA and is
not very old then the cylinder size is 8193024 bytes.  When LBA is not
used the cylinders are smaller.

Anton Zinoviev

Reply to: