Vgextend fails

[Adam Winebaugh]

Anyone,
I have a disk I am attempting to add to vg01. Here we go:
# vgextend /dev/vg01 /dev/dsk/c20t4d5
vgextend: Warning: Max_PE_per_PV for the volume group (4342) too small for this PV (6399).
Using only 4342 PEs from this physical volume.
vgextend: Couldn't install the physical volume "/dev/dsk/c20t4d5".
The path does not specify a valid physical volume.

Odd it says that it is not a "valid" physical volume. I know it is. But I also notice that the maxpe per pv is off so just to test I ran apvdisplay with no arguements just to see if it would work:
# pvcreate /dev/rdsk/c20t4d5
pvcreate: The physical volume already belongs to a volume group
I run a vgdisplay on all vg's to find the disk and it does not show on ANY of the VG's. So what vg is this error referencing?

Thoughts anyone? Should I use the -f and -s options on the pvcreate and specify the size? then perhaps it will force it into submission and I can vgextend it?

[Mel Burslan]

you definitely will need -f switch on the pvcreate comand to overwrite the older PV and VG information on this disk. I am not sure about the -s option though as I have never used it. What you are experiencing is a planning problem from the beginning. When you are creationg volume groups, rather than going with the default PE_per_PV value, you should specify a larger number that will accomodate larger disk drives.

There is a quasi-unsupported vgchange utility from your HP account representative, which may or may not help you change this PE_per_PV setting on your volume group but the anectodal evidence says, its success is about 50% unless HP released a newer version. It is worth to investigate as an option. Otherwise, your best bet is tot recreate the volume group with larger PE_per_PV value and restore your data into it.

[Adam Winebaugh]

Mel,
You say old VG and PV data on the disk? This disk has never been utilized before. So, I am a little confused I guess as to why it would have vg/pv information on it.Also, yes, I agree about the max pe per pv but to little to late ya know. I don't want to risk anything.

Any other thoughts on a cause?

[Adam Winebaugh]

Ok, I used the -f option and the pvcreate passed. so did the vgextend. I figured I would have to use the -f option, but always better to have another opinion, thanks Mel.

I may try the pv/vgchange commands and see what happens.

[Adam Winebaugh]

Thanks again.

[Mel Burslan]

even if you think the disk drive was never used, it may be a refurbished unit given to you by your reseller, or it may be recyled from another department. I don't know your exact situation but I have seen enough many "new" disks, which were previously used under LVM, such that they had vg information located on their header sections. So, -f was a no brainer.