vinum unable to write config on 5.4

jason kawaja kawaja at
Wed Aug 3 19:04:16 GMT 2005

On Wed, 3 Aug 2005, Andrew Belashov wrote:

> jason kawaja wrote:


>> vinum -> resetconfig
>>  WARNING!  This command will completely wipe out your vinum configuration.
>>  All data will be lost.  If you really want to do this, enter the text
>>  Enter text -> NO FUTURE
>> Can't find vinum config: Inappropriate ioctl for device
>> if i create a vinum partition on a 5.3-release then the 5.4-release can
>> read it fine, but is unable to make changes to the vinum config.
>> ideas on changes between the two (5.3 - 5.4) that may have induced this
>> behavior?
> Can be it will help:

that put me on the right path, thanks.  im gonna conclude that in 5.4 
the original vinum on sparc64 is unable to write configuration 
information to disk, that appears to be the only problem (in my case).

shown below are the steps i took to remove a vinum configuration and 
then create it (both steps on 5.4-RELEASE-p6) once the vinum.ko was 
unloaded and geom_vinum.ko loaded :

resetconfig steps

gvinum rm stripe.p0.s0
gvinum rm stripe.p0.s1
gvinum rm d1
gvinum rm d2
gvinum rm stripe.p0
(error displayed on above command, ignored)
gvinum rm stripe
(error displayed on above command, ignored)
gvinum saveconfig
(maybe just reload module, but i rebooted)

create steps

gvinum create gvinum.cfg

# drive d1 device /dev/da1d
# drive d2 device /dev/da2d
# volume stripe
#  plex org striped 4096k
#   (ive done my own testing, dont talk to me about 4096k)
#   sd drive d1
#   sd drive d2 
#   (normally use "size 0" for subdisks but gvinum didnt like that)

gvinum saveconfig
newfs -U /dev/gvinum/stripe
fsck -t ufs /dev/gvinum/stripe
(above is a preference, i like to fsck new partitions)

other changes 

- inside /boot/loader.conf add following line


- may not be needed: inside /etc/rc.d/vinum alter the following variable

start_cmd="gvinum start"

- add entry in /etc/fstab

/var/log/vinum_history is no longer updated, guess that's a gvinum issue 
and not sparc64 specific though.  next round of testing is to create a 
mirror, break it abnormally and attempt to attach a new device to 
establish the mirror again.

Jason Kawaja 2-4568

More information about the freebsd-sparc64 mailing list