Unable to extend a VG 1

[KPKIND]

Hi,

I have had a problem extending rootvg. I was trying to replace an internal disk on AIX 5.2, as a part of which I have unmirrored the VG and then deleted the disk(rmdev -dl hdisk0). I have also verified by doing lspv hdisk0 which did return saying "Unable to find hdisk0 in config database". Now I have got the disk replaced, after the system has come back for some reason I was unable to add the disk to the Volume group.The error I was getting was as below

0516-008 extendvg: LVM system call returned an unknown
error code (-271).
0516-792 extendvg: Unable to extend volume group

I have searched the whole web but could not find any solution, except for one other guy who has faced this problem with no solutions from any of the guys. I have almost struggled for an hour and a half, finally when I have looked in to the ODM for rootvg there ware two pvid's which are a part of the rootvg, but in real time I would expect only one pvid in the rootvg as the rootvg currently was running with one disk. So I have got rid of this pvid which has fixed the problem. I have executed
"synclvodm -P -v rootvg" which has got the right informaiton from the LVCB of the disks and updated the ODM and that fixed the problem.

Thought it would be useful for someone who might face such problem, hence I am posting this.

Cheers
KPKIND

 

[mrn]

Couple of (maybe daft) questions

Did you run cfgmgr?
Did you re-activate the scsi slot?

Mike

"A foolproof method for sculpting an elephant: first, get a huge block of marble, then you chip away everything that doesn't look like an elephant."

http://www.airport-parking-site.co.uk/

 

[KPKIND]

After the disk was removed from the rootvg, I have done a rmdev and then had to restart the server as the disk I am replacing was an internal disk. So I am sure the system has run cfgmgr during the startup.

Coming to the reactivation of scsi slot, I have put the new disk in the same slot where I had the old disk.

Cheers
KPKIND

 

[mrn]

So if you issue a lspv can you see the disk?

If you remove the hdisk0 to replace it then you should go to smitty devices/scsi slots and reactivate the
slot; when you remove a drive the slot it is disabled and the new hdisk0 you see recognized in your system is not the new but the old one.


Mike

"A foolproof method for sculpting an elephant: first, get a huge block of marble, then you chip away everything that doesn't look like an elephant."

http://www.airport-parking-site.co.uk/

 

[p5wizard]

@Mike, KPKIND already solved his problem (re-read second part of initial post) - no need to spend more time&effort...

@KPKIND, perhaps you should have entered this as a "TIP" ?

HTH,

p5wizard

 

[mrn]

Doh!

Mike

"A foolproof method for sculpting an elephant: first, get a huge block of marble, then you chip away everything that doesn't look like an elephant."

http://www.airport-parking-site.co.uk/

 

[KPKIND]

Hi p5wizard - I should have done that.

But anyway Mike, I wanted to know what you were talking about. I could not see the disk when I have issued a lspv. You said we need to go into smitty devices then scsi slots, but I could not see any such option in the smit panels and I have never heard of this option of reactivating scsi slots. Can you give me the exact path for this option please.

TIA
KPKIND

 

[p5wizard]

What I think happened was:

Your reducevg/rmdev did not completely erase hdisk0's PV info from ODM. (Did you do the reducevg at all? You didn't mention it in the original post - but then how would you be able to rmdev a disk if it is still a VG member???)
And then when your system booted and cfgmgr-ed the replaced disk, it was unable to assign a new PVID to the newly found hdisk0, because in ODM, there was still some info pertaining to the old hdisk0...

As to scsi slot (re)activation - that is new to me too.


HTH,

p5wizard

 

[mrn]

From memory it's part of diag (But I'll double check when I'm in front of a machine); also take a look at the lsslot command

lsslot -c pci

Mike

"A foolproof method for sculpting an elephant: first, get a huge block of marble, then you chip away everything that doesn't look like an elephant."

http://www.airport-parking-site.co.uk/

 

[p5wizard]

Mike, pci slot hot plug and scsi slot hot plug operations are two different things altogether. Granted, they are found in diag Task Selection under the same menu heading...

But I understood from KPKIND that the disk was replaced with machine down, so no hot plug scsi anyway...


HTH,

p5wizard

 

[KPKIND]

Hi p5wizard,

My reducevg command has worked perfectly, then rmdev has worked perfectly. I am saying this because before I have shutdown the box I have executed the command lspv hdisk0, which has returned "Unable to find hdisk0.....". I am 100% sure of this. In the light of this problem when I have done the same task on another disk again in the rootvg - on the same system, I have made sure there was no entry in the ODM for the disk that I have removed (This time it is hdisk1. I have checked through the ODM using odmget command and even that did not return any entry. Somehow it has picked it up during the reboot), but the problem still reoccured.

I once again confirm that the disk replaced was a scsi disk which was not hot pluggable.

TIA
KPKIND

 

[ralisp]

Thanks for the tip! I had the same problem and this was the only post that had the fix.

 

[DukeSSD]

KPKIND, please post the oslevel -r for the system, there are all sorts of LVM/ODM problems on 5.2.
While we are on the topic of 5.2 problems there are also a lot of changes in ML 3, 4, 5 that will affect systems with old firmware - update it before applying any software updates, as IBM keep telling us.

 

[J1gh2]

Hi Guys

I would also appreciate your help with an issue where one internal SCSI disk (hdisk1) has dropped off from rootvg and I am trying to get it back to mirror rootvg.

I ran into problems when I updated the effective configuration of a SAN switch which resulted into a device going into an "incorrect state" (with numerous FCP_ARRAY_ERR6) and hdisk1 dropping off from rootvg. This affected all the devices and what I removed all the devices and ran cfgmgr again, the other devices have come back in a consistent state but the internal scsi disk still missing.

hdisk1 is now a "fcparray Disk Array Device" so I think the odm reference to the original hdisk1 was removed for this name to be given to a new device. I am trying to get the other scsi disk to show up so that I can mirror rootvg.

I would appreciate any tips to get the other internal disk to be configured and visible. below is the list and status of the scsi adapters:

scsi0 Available 06-08-00 PCI-X Dual Channel Ultra320 SCSI Adapter bus
scsi1 Available 06-08-01 PCI-X Dual Channel Ultra320 SCSI Adapter bus
scsi2 Defined 09-08-00 PCI-X Dual Channel Ultra320 SCSI Adapter bus
scsi3 Defined 09-08-01 PCI-X Dual Channel Ultra320 SCSI Adapter bus

I am using AIX 5.3.

Thanks