Oops, sorry - I forgot about this one.
On Thu, Apr 10, 2003 at 11:09:05PM -0700, Vinh Ly wrote:
> disto = forRH-8.0-SGI-XFS-1.2.0-v4.iso, RedHat patches current as of last
> Monday
> kernel = 2.5.66 + linux-2.5.66-xfs-2003-03-30.patch
> md = mdadm 1.2
> raid level = 0 and 5 (4 SCSI disks)
> xfsprogs = 2.3.5 and 2.3.9
>
> When I run the command "mkfs.xfs -f /dev/md0", I get this error:
>
> MD array /dev/md0 not in clean state
>
> Version 2.3.5 and 2.3.9 give the same error.
>
> It looks like the way it checks for the
> md state is different. Here's the snipet from libdisk/md.c:
>
> /* Check state */
> if (md.state) {
> fprintf(stderr, _("MD array %s not in clean state\n"),
> dfile);
> exit(1);
> }
>
> In the mdadm 1.2 source (Examine.c), they have this snipet:
>
> printf(" State : %s, %serrors\n",
> (super.state&(1<<MD_SB_CLEAN))?"clean":"dirty",
> (super.state&(1<<MD_SB_ERRORS))?"":"no-");
>
> and super is defined as mdp_super_t (called md_array_info in xfsprogs).
> And MD_SB_CLEAN=0 and MD_SB_ERRORS=1.
>
> The output from "mdadm --examine /dev/md0" and "cat /etc/mdstat" look fine.
>
> The mkfs command works fine on 2.4.20 w/ the xfs patch from
> snapshot-xfs-2.4.20-2003-03-19_04:55_UTC.
>
> Any ideas? Am I supposed to use LVM instead in 2.5.x?
No, this looks like a mkfs bug. I will checkin the fix shortly,
it seems fairly simple from a quick look at it.
thanks.
--
Nathan
|