Daniel Moore wrote:
>
> => Maybe more people with 2.4.6-pre1-xfs or newer can comment on whether
> => xfs_repair with the -n option fails. It seems like what is broken is
> => that it says you can't run it on a mounted partition, even if it is -n
> => and won't write. If not, it seems that the call that responds with
> => "fatal error -- couldn't initialize XFS library" is saying that some
> => sort of internal function failed...don't know, I'm not familiar with it.
>
> I came in half way through on this thread...
>
> Why would you want to run xfs_repair (with -n or otherwise) on a mounted
> partition? The FS is almost certain to be inconsistent given that xfs_repair
> doesn't look at the log.
I had hoped it would at least comment on its thought about what is or
isn't consistent. I can't unmount the root partition, and so far am
unable to create any kind of boot medium other than a removable IDE hard
drive (the ultimate floppy).
The slackware style boot floppies are not working for me either, partly
it seems because I have no ability to format a floppy to the size 1600.
I have 1660 and 1680, but no device file for 1600 even, and this is
apparently required to be exact. I absolutely require a boot image that
supports both XFS and SCSI aic7xxx, so I have to use larger than 1.44
MB. I have used yard to create all kinds of variations, but under no
circumstances is it possible for me to create an image less than 2.3 MB,
and no floppy will go that far (from a 1.44 MB drive). Therefore I am
experimenting with cd rom images, which must be 2.88 MB or smaller. My
problem here is I must create it on linux, and transfer it to a
braindead win 2k machine that does not have the aspi layer mandatory for
writing images directly, and apparently there is no support for this on
win 2k without a miracle.
So I wanted a means to at least do *some* kind of test to see if things
have gone horribly wrong while it is still mounted, read-only. After
rebooting and having it forget all settings for all pci devices, I felt
there was a need.
D. Stimits, stimits@xxxxxxxxxx
>
> I suspect you're encountering a feature here.
>
> -----------------------------------------------------
> Daniel Moore dxm@xxxxxxx
> R&D Software Engineer Phone: +61-3-98348209
> SGI Performance Tools Group Fax: +61-3-98132378
> -----------------------------------------------------
|