Mirror disk

[james1964]

We have a standby server. We were trying to mirror the disk and after the lvlnboot command we got an error message that said Warning cannot determine all physical Volumes on which mirrored copies of the logical volumes are located. There is more, but then it also says vgcbackup: /etc/lvmtab is out of date with the running kernal. Kernal indicates 3 disk for /dev/vg00 lvmtab has 2 disk.

We are able to boot of the 1 disk, but when I do a bdf command it does not show lvol1 (/stand) This is seen in the fstab file. When I look at lvmtab it shows the vg00 vol group and both physical disk. But I can't do any lvln command without error.

I have tried vgscan and vgimport, but they both show all 3 vols exist. If I try and mount /stand it says it does not exist.

This is confusing to me. The server should not boot if /stand does not exist. Any Ideas?

[Steven E. Protter]

Shalom,

Perhaps the /stand filesystem is corrupt. Try running fsck on it.

SEP

[Geoff Wild]

Can you post a
vgdisplay -v vg00

dmesg

Is there anything in /stand?

Did you try a
vgreduce -f

Rgds...Geoff

[Tim Nelson]

If you have just implemented the attempt to mirror I would suggest to start over.
Reduce the lvols and devices that you added and get back to a working vg with no errors.

Reduce back to working order and then re-do your steps to mirror.

Somewhere in here you lost a device.

[james1964]

This server is not on the network. Since it is a standby it has the same name as a live server.

Out put from vgdisplay is as follows

VG Name /dev/vg00
VG write access Read/Write
VG Status Available
Max LV 255
Cur LV 3
Open LV 3
Max PV 16
Cur PV 2
Act PV 1
Max PE per PV 4384
VGDA 2
PE Size (mbytes) 16
Total PE 4374
Alloc PE 4085
Free PE 289
Total PVG 0
Total Spare PVs 0
Total Spare PVs in use 0

--- Logical Volumes ---
LV Name /dev/vg00/lvol1
LV Status Available/Stale
LV Size (Mbytes) 304
Current LE 128
Allocated PE 256
Used PV 1

--- Physical Volumes---
PV Name /dev/dsk/c1t15d0
PV Name /dev/dsk/c3t15d0 Alternate Disk
PV Status Available
Total PE 289
Free PE 289
Autoswitch On

I have not run the fsck yet. I want to learn more about the command before I run it.

[Torsten.]

Why did you only post parts of the output?

OK, I assume you have an A-400 or 500 with two internal disk, but they are not mirrored.

Look:

Act PV 1 <== 1 disk
LV Name /dev/vg00/lvol1
LV Status Available/Stale <== "stale" = BAD
LV Size (Mbytes) 304
Current LE 128
Allocated PE 256
Used PV 1 <== only 1 disk used
..
PV Name /dev/dsk/c3t15d0 Alternate Disk <== I know only "Alternate link"


Please don't send modified output like this!


A "lvdisplay -v" for this lvol would be nice, but I really guess you mirrored the LVOL to the __same__ disk for any reason.

[sdass]

Provide the ioscan -fnC disk,diskinfo for all available disk, lvdisplay -v /dev/vg00/lvol1 output

[james1964]

Thank you all for teaching more about how to get info from the drives. The mirror never finished due to the error in the middle of it. The /stand volume I believe was corrupt. We decided to start over using the 2nd disk from the live server, and remirroring the live server with a blank disk.

I wrote down everything I saw from the output and typed it in. If there was suppose to be more I did not see it.

Thanks again for all your support

[Torsten.]

You should use a terminal emulator like putty or hyperterminal and enable logging - this helps finding errors later.

[james1964]

Normally I would use putty and cut and paste, but this server is not on the network. The NIC is not plugged in at all. I am working from a console dumb terminal from the server room.

I tried pulling a drive from the live server, but the live server has 1 CPU and the standby server has 2 CPU's. I can't seem to get it to boot. Is there something in the kernal that would not allow a 1 CPU drive not to boot on a 2CPU server? Is there anything I can do about this?

[Torsten.]

Same system model, same disk position?

try to use a laptop/PC and a null-modem cable to ba able to capture the output.

[Geoff Wild]

Yes it is possible - either via NFS or Cluster File System.

Course, with CFS, you need ServiceGuard:

http://docs.hp.com/en/T1859-90048/ch03s11.html


Rgds...Geoff

[james1964]

Excellant Idea. I did that before with the laptop. Should have thought of that. Yes same server model (RP2470) same disk position.

How does NFS help with booting a 2 cpu server with a 1 cpu disk?

[Torsten. · Accepted solution]

I guess Geoff posted to a wrong thread.


1 or 2 or 100 CPU - this makes no difference.

Did you pull the disk while the server was up and running???

The filesystem may be bad now - you really need to post what you see on the screen.

[james1964]

When I did that I had a drive from the shelf that was mirrored from the 1 CPU system when it first went live. The server was down when I pulled the drive. The error I got was from the GSP. Sorry I did not copy it down.

To move forward, I still had 1 drive from the first senerio build. I put it in bay 2 and mirrored to blank in drive bay 1. This is working. It is still syncing. (Was not aware that the drive bay matter so much). It seems if you try and mirror a drive that is placed in the wrong bay it comes back with an error on any LVLNBOOT command.

Thank you Torsten for the help and the knowledge that the 1 cpu drive should work with 2 CPU server. If I run into anymore problems, I will try to make sure I hook up a laptop and cut and paste errors and command responses.

[Geoff Wild]

Ah yes - sorry - I posted in the wrong thread....

Rgds...Geoff