Forum Discussion

gmathew's avatar
gmathew
Level 3
15 years ago

Cannot remove last disk group configuration copy

Hi, I have a diskgroup with 6 EMC SAN disks in it. I got 6 new SAN storage and added it to the same disk group and mirrored volume. This host is running on centos 4.8. After mirroing I ha...
business continuity
Storage Foundation
  • gmathew's avatar
    gmathew
    15 years ago
    Thanks Everyone. I could resolve the issue. It was not the problem with the disk. There seems to be a bug or a issue with VxVM 4.1mp3 on Linux, wherein if we add the disk to a dg and not reboot the linux box, we cannot remove the unused disk because the config copies doesnt get created on the new disks after we add it to the diskgroup.

    Rebooting the linux box solved the issue. The config copies are created on 5 of the new disks after the reboot.

    Thanks again for all your valuable help!

g_lee's avatar
g_lee
Level 6
15 years ago
at the risk of stating the obvious, that dg doesn't look healthy!

eg: config copies on the 6 "new" disks (emcpower[q-v]) state=iofailed, Error: error=Disk read failure
That state tends to be seen when the dg has lost access to the disk and has gone to dgdisabled state, eg: in this technote:

TN 229701: What to do when "vxdisk list" shows status of 'online dgdisabled'
http://seer.entsupport.symantec.com/docs/229701.htm

The dg probably hasn't gone into a disabled state (yet) as you still have a good copy on the "old" disk emcpowerf (dumpdg06) -- are there any issues on the volumes themselves?

Would you be able to provide the output of vxprint -qthrg <dg> as Dev requested? ie:
# vxprint -qhtrg dumpdg

From the currently available output it looks like the system has lost access to the disk at some point (possibly temporarily, or at least long enough for it to not be able to get to its config copy), OR something has happened to the label/header/portion on the disk where the config copy was stored.

The disks and dg are still showing online, which is why it would be good to check the vxprint output before proceeding with anything to see what is going on with the actual volumes - also, are the volumes in use at the moment?