Solaris

Hi All
BAsed on the below I would like to verifu two things
(1) The udnerlying mirroris for '/mnt' na dit onlcy contaisne 1 sub-mirror, with one sliceon is one disk and hence, data loss on the mount point (the mount point, '/mnt' is backed up)
(2) the Procedure for renewal

Code:

# df -kh /mnt
/dev/md/dsk/d6          33G   1.2G    32G     4%    /mnt


# metastat d6
d6: Mirror
    Submirror 1: d26
      State: Needs maintenance
    Pass: 1
    Read option: roundrobin (default)
    Write option: parallel (default)
    Size: 71127180 blocks (33 GB)

d26: Submirror of d6
    State: Needs maintenance
    Invoke: after replacing "Maintenance" components:
                metareplace d6 c1t8d0s0 <new device>
    Size: 71127180 blocks (33 GB)
    Stripe 0:
        Device     Start Block  Dbase        State Reloc Hot Spare
        c1t8d0s0          0     No      Last Erred   Yes

1) Obtain a copy of the vtoc of the disk either from the prtvtoc command or from an old explorer report
2) Take a copy of the SVM configuration " metastat -p"
3)
# metadetach -f d6 d26
# metaclear -f d6
**Power Down to Single Suer Mode
4) Un-mount the file system being used on the disk
umount /dev/rdsk/c1t8d0s0
*** if it is '/tmp' will it handg ??? ***
6) Replace the disk using cfgadm -c disconnect / cfgadm -c connect or luxadm replace - depending on the disk type
7) Put a new vtoc on the replacement disk
fdisk
8) Metainint the SVM SVM volume needed - See the metastat -p output
9) Newfs and mount the filesystem -
10) Add back any removed metadb's ... metadb -a -c X /dev/rdsk/..... X=number of copies needed

11) Boot the system to multi user if the system was shut down to single user

Moderator's Comments:

Please use code tags to improve readability (and as required by forum rules)! And, please watch your orthography! While you may save a few seconds when typing carelessly, ALL readers will have to fight to understand above scrawl.

Do not forget to firstumount -f /mnt!
And check if it holds a metadb copy:

Code:

metadb

and eventually delete an affected one with metadb -d ...
Normally it should not be necessary to shutdown. But there might be bugs in Solaris LVM that make it advisable.(?)

What is special about /mnt that this operation cannot be done online ?
Are other slices of that same failed/falling disk being used for other purposes ?

You should post complete output of meta* commands, with mount output to see the entire disk layout.
That way help could be much more precise, and perhaps might not involve rebooting at all.

Hope that clears things out
Regards
Peasant.