More than 255 raid devices

Neil Brown neilb at suse.de
Mon Sep 4 03:10:14 UTC 2006


On Friday September 1, info at contextshift.co.uk wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Fri, Sep 01, 2006 at 05:11:43PM +0200, martin f krafft wrote:
> > first off: please either subscribe to the list or indicate that you
> > want to be CCd according to Debian list policy. No biggie...
> 
> Oops, sorry, Yeah, can I be cc'd in on the replies please?

I'd much rather address the mail to you, and Cc the list.  This idea
of replying only to the list never made sense to me, but I guess
everyone is different....

> 
> > also sprach J D Freeman <info at contextshift.co.uk> [2006.09.01.1522 +0200]:
> > > brains:~# mknod /dev/1000 b 9 1000
> > > brains:~# ls -al /dev/md1000
> > > brw-r--r-- 1 root root 9, 1000 2006-09-01 13:58 /dev/md1000
> > > brains:~# mdadm -C /dev/md1000 -l 1 -n 3 /dev/vg01/testvg /dev/vg02/testvg missing
> > > mdadm: error opening /dev/md1000: No such device or address
> > 
> > Does this generate any output in dmesg?
> 
> Absolutely nothing, logged anywhere. dmesg has no entries for the past
> few days.

No, there wouldn't be...
The 256 is a hard codes limit in include/linux/raid/md_k.h
#define MAX_MD_DEVS  256	/* Max number of md dev */

It could safely be increased to about 8192 I think and probably made
to go a little higher with a bit of care.
But I'm afraid it will need a recompile of the kernel.

I'll see about fixing this properly for 2.6.19.

NeilBrown



More information about the pkg-mdadm-devel mailing list