Failed root disk and can't get a replacement disk

[bahr]

I'm in a kind of ugly situation. I have a solaris server running solaris 9 and veritas volume manager 5.0. The root disk is mirrored. The server was rebooted last week and couldn't boot off the root disk. I was able to successfully boot off the mirror, and can see in vxdisk list that the root disk is reporting as online failing. The problem is that my management is dickering over our Sun contract, which has expired, and they won't authorize me to get a replacement disk. So I'm running off the mirror for I don't know how long.

questions -

What bad things will happen if I continue running off the mirror? Do I need to turn the mirror into a root disk and run without a mirror? I don't think the mirror is encapsulated, do I need to encapsulate it if I turn it into a root disk? And can anyone point me towards documentation that would walk me through doing this?

[Gaurav_S · Accepted solution]

Hello,

yup, situation looks ugly.... however to answer your queries...  (I presume when u say that you have booted through mirror.... you mean booted through slices of mirror disk)

a) I don't see any downside of keeping mirror running
b) As you say that rootdisk has got a online:failing flag, failing flag usually indicates a temporary write failure, if you check in "iostat -En" output, do you see many hard errors on the actual root disk ? it is quite possible that disk isn't fully gone bad & has few bad sectors... you an use "format" utility of solaris to "analyze" your root disk & mark off bad sectors

c) Once you are done with step b, if you find too many bad sectors or too many io errors, then you don't have a choice to mirror, then you have to survive on one disk only.... now it doesn't really matter if u keep it encapsulated or unencapsulated, main purpose of encapsulation comes with mirroring, as you don't have mirroring.. no use of encapsulation...

to conclude, only hope looks to be how much ur rootdisk has gone bad, if it is good to survive we can think to mirror again....


Gaurav

[sunshine_2]

this is the correct time to create a flar of the OS and keep it aside just in case the other disk goes bad too ..

the command is

# flarcreate

command to remove online failing flag is

# vxedit -g <disk_group> set failing=off <sub_disk>

[wzis]

Please post:
1. vxdisk list
2. vxprint -htg rootdg
3. df -k /

[Marianne]

Surely the mirror disk must have been encapsulated or initialized before it was added to rootdg?
New disk are normally initialized when placed under VxVM control - this will wipe existing partitioning (and all data).
Disks containing data (like boot disk) are encapsulated in order to preserve existing data.

To confirm you can also run
prtvtoc /dev/rdsk/c#t#d0s2
Confirm presence of tag 14 and 15.