Frank:

Maybe I muddled the waters by not explaining our whole upgrade process and here 
it is.

We are performing a Live Upgrade on a machine that has the following metadevice 
configuration:

# Mirror of /
d500 -m d400 d700 1
d400 1 1 c0t0d0s0
d700 1 1 c2t0d0s0

# Mirror of swap
d501 -m d401 d701 1
d401 1 1 c0t0d0s1
d701 1 1 c2t0d0s1

# Mirrors of /var
d503 -m d403 d703 1
d403 1 1 c0t0d0s3
d703 1 1 c2t0d0s3

# Mirror of /export/home
d507 -m d407 d707 1
d407 1 1 c0t0d0s7
d707 1 1 c2t0d0s7

and all the file systems have UFS logging turned on.

Here's our upgrade process:

* metadetach d700, d701, d703 & d707 metadevices without doing a lockfs -w 
because we don't care about the data on c2t0d0.
* create the new BE by calling lucreate with -s option so that the new BE does 
not get populated.
* luupgrade the new BE using flar so that there is a completely new image of 
the OS in the new BE i.e. all my c2t0d0 slices have the new OS. 
* luactivate the new BE boot into it using shutdown or init. This means we are 
booting off the c2t0d0 slices.
* after successful boot into the new BE (c2t0d0), sometimes we need to mount 
the old BE (c0t0d0) slices for reference purposes.

So a 

# mount /dev/dsk/c0t0d0s0 /mnt 

causes the system to panic with a "freeing free block" message.

Is this happening because of logging in the old BE?

Thanks,
-- osdummy
This message posted from opensolaris.org
_______________________________________________
ufs-discuss mailing list
[email protected]

Reply via email to