On Fri, 21 Feb 2003, James Rich wrote:
> 2.4.20 - though my cvs isn't as up to date as I thought it was. Was
> BLKGETSIZE64 broken anytime during 2.4.20?
Nope, much earlier.
> I'm just wondering if my log partitions are big enough and if the size
> parameter is not specified does the log occupy the entire partition?
That's right, the external log size defaults to the size of the log device
you point at.
If the device is smaller than what you request, you should get something like:
mkfs.xfs -f -l logdev=/dev/sda2,size=10000b /dev/sda1
size 10000b specified for log subvolume is too large, maximum is 4016 blocks
If my test device is large enough, then the command above succeeds for me.
I thought that you had to specify a block size before you could use a
"b" unit to specify any other size, but it looks like that's not
causing an error...
In any case, you were specifying 10000b, which is using (I guess) 4k blocks,
or a 40960000-byte log. You said /proc/partitions showed 16033 512-byte
blocks for sdb6, or 8208896 bytes - smaller than what you asked for.
So that might be why it's failing, but I don't know why it's failing the way
it is... I'd probably need to start gdb'ing or printf'ing my way through
mkfs to see what's going on.
-Eric
|