07-29-2013 08:39 PM
Dear All,
Kindly I have the following system
Solaris 8 with Veritas Volume manager 3.5
I had two disk three disks failing
two of them in a volume tdisk4 and the third one was part of another volume named tdisk5
the thing is that I successfully replaced the three disks and now the systems saw them and I can manage them but when I try to put the new discovered luns under VxVm control it failed
I issue the command
vxdiskconfig
or
vxdctl enable
then when I issue the command
vxdisk list
It shows that they have error status
they are given a name T3_0 , T30_1 , T30_2 , but with error status
What could be the possible cause for that
vxdisk list T30_0 for example it shows that the error is " Device path is not valid"
However the lun is seen successsfully from the Solaris OS prompt and I can edit its slices , .. etc
Thanks for your help in advance
Solved! Go to Solution.
07-31-2013 12:45 PM
Do I need to do anything regarding the format ? of any of the disks ?
Newly added disks/luns need to be labeled using the Solaris 'format' menu.
07-29-2013 08:43 PM
Kindly fine below the output of the some commands to give you a further information
07-29-2013 09:36 PM
Hi,
Can u paste the "vxdisk -eo alldgs list" output ?
Gaurav D
07-29-2013 10:16 PM
I'm not now in front of the server but I remember the output of the command
vxdisk -e list
which added the last column of the path of each lun and the path was correct
also the command of vxdisk list T30_0
for example shows the two redundant paths for the lun and both were enabled
eventhough the error says device path is not valid
!!!!
hope this answer helps you and if you still need the output of the command above I will past its complete output tomorrow :)
Thanks :)
07-29-2013 10:17 PM
You need to initialize the new disks and then add them to the respective diskgroups:
vxdisksetup -i T30_0
vxdg -g t3dg -k adddisk t3dg02=T30_0
Since your volumes have no redundancy, you will need to restore from backup.
07-31-2013 02:04 AM
Dear Marianne
I run the command
vxdisksetup -i T30_0
but it says
Error Device T30_0 online failed
Device path not valid
do you have any clue why ?
Thanks
07-31-2013 02:21 AM
can you post the output
# vxdisk list T30_0
07-31-2013 03:26 AM
Hi Mahmoud,
I guess the OS enumerated the new LUNs with the old device ID's or your OS device tree is not cleaned up properly.
-Cleanup Solaris device tree
# devfsadm -Cv
# vxddladm stop eventsource
-Cleanup powerpath device tree (Only needed if EMC Power Path is installed and configured)
# powercf -q
# powermt config
# powermt check
-Cleanup VxVM device tree
# mv /etc/vx/disk.info /etc/vx/disk.info.old
# mv /etc/vx/array.info /etc/vx/array.info.old
# rm /dev/vx/dmp/*
# rm /dev/vx/rdmp/*
# vxdctl enable
# vxddladm start eventsource
Then try again to initialize the disks and add them to the respective diskgroups.
Another option would be to reboot the system, this will also clean up the device tree
07-31-2013 03:49 AM
07-31-2013 04:52 AM
Please apologize, I missed the 3.5 version.
The persistent feature which uses the disk.info and array.infow as introduced in 4.1, hence you can't find the files in 3.5.
vxddladm is a command that resides in the sbin directory.
Can you try the above procedure without the renaming of the *.info files?
07-31-2013 05:05 AM
vxddladm doesn't work with the option given above
vxddladm stop eventsource is not working
It seems to be a wrong option
It's worth mentioning now that T30_0 suddenly work and now it's online :D
What do you think happened here ?
07-31-2013 05:05 AM
Do I need to do anything regarding the format ? of any of the disks ?
07-31-2013 08:03 AM
If you ran all the other steps apart from the vxddladm command, then I assume your device tree is clean and the dmp nodes are referencing correct OS devices.
Sorry for wrong option, but it's been a couple years I touched a 3.5 installation ;)
As the disks are all sliced, the new disks format should also be sliced, if that is your question.
If this is a cluster, please also make sure that the devices are correctly discovered on all nodes.
07-31-2013 12:45 PM
Do I need to do anything regarding the format ? of any of the disks ?
Newly added disks/luns need to be labeled using the Solaris 'format' menu.
07-31-2013 05:45 PM
Dear All thanks all for your help
Finally I solved the problem
each lun has two paths
I just found that one of the paths the disk is formatted by the veritas volume manager and the other not
so I prtvtoc the vtoc from this one labled by VxVm and applied to the other path of each lun , hence it was changed in the vxdisk list from error status to online invalid
from this point I continued successfully to initialize the disk and add it back to all the defected disk groups ,.... etc until I finally recover the volumes and all now are enabled active
Of course I lost all the data and now I'm restoring the data from the backup the customer have using Veritas Netbakup 5.1 which is a very old version
Wish me luck restoring the backup because it's a very old data and and very old backup solution as you can see :)
Thanks again so much for your help