Add disk to Volume Group 1

[djml]

I have 2 nodes with AIX 5.3 and HACMP and i need to add more disks to a volume group:

Node 1:
root@Node01$ lsvg -o
dbvg
dbvgextend
rootvg
root@Node01$ lspv
hdisk0 00012b3a374a6fa0 rootvg active
hdisk1 00012b3a19bdf151 rootvg active
hdisk2 00012b4a48c3cdba wasvg
hdisk3 00012b4a48c438ab dbvg active
hdisk4 00012b4a48c45cea dbvg active
hdisk5 00012b4a48c4631f dbvg active
hdisk6 00012b4a48c46954 dbvg active
hdisk7 00012b4a48c46fb3 dbvg active
hdisk8 00012b4a48c47605 dbvg active
hdisk9 00012b4a48c47c40 dbvg active
hdisk10 00012b4a48c48277 dbvg active
hdisk11 00012b4a48c3e1b2 wasvg
hdisk12 00012b3a456a36a4 dbvgextend active
hdisk13 00012b3a456adfc6 dbvgextend active
hdisk14 00012b3a85df3ad3 None
hdisk15 00012b3a85dfaf3a None
hdisk16 00012b3a874f1ecd None
hdisk17 00012b3a874f69f1 None


Node 2:

root@webaix02$ lspv
hdisk0 00012b4a3c26ca93 rootvg active
hdisk1 00012b4a2a366b81 rootvg active
hdisk2 00012b4a48c3cdba wasvg
active
hdisk3 00012b4a48c438ab dbvg
hdisk4 00012b4a48c45cea dbvg
hdisk5 00012b4a48c4631f dbvg
hdisk6 00012b4a48c46954 dbvg
hdisk7 00012b4a48c46fb3 dbvg
hdisk8 00012b4a48c47605 dbvg
hdisk9 00012b4a48c47c40 dbvg
hdisk10 00012b4a48c48277 dbvg
hdisk11 00012b4a48c3e1b2 wasvg
active
hdisk12 00012b3a456a36a4 dbvgextend
hdisk13 00012b3a456adfc6 dbvgextend
hdisk14 00012b3a874f1ecd None
hdisk15 00012b3a874f69f1 None
hdisk16 00012b3a85df3ad3 None
hdisk17 00012b3a85dfaf3a None

I want to add the disks hdisk14 and hdisk15 to the VG - dbgv and the disks hdisk16 and hdisk17 to the volume group dbvgextend

 

[samirm]

What is the lsvg -o output for Node 2 ?

 

[djml]

The Node 2 have this output for the lsvg -o:

root@Node02$ lsvg -o
wasvg
rootvg

 

[samirm]

Just show lsvg output of both the nodes ...

I want to know how failover has been configured with respect to VG. It may looks like "lsvg" output
for Node-2 :-

wasvg
rootvg
dbvg
dbvgextend

And for Node-1 :-

rootvg
dbvg
dbvgextend
wasvg

If you want to extend the VG dbvg and dbvgextend,
just do in Node 1 :-

extendvg dbvg hdisk14
extendvg dbvg hdisk15

extendvg dbvgextend hdisk16
extendvg dbvgextend hdisk17

But prior to do this you may have to stop the cluster service for both the nodes by # smitty clstop

Now go to Node-2. do following :-

exportvg dbvg
exportvg dbvgextend

importvg -y dbvg hdisk7 -->> It can be any disk which is associated with dbvg for Node 2

importvg -y dbvgextend hdisk12 -->> it can be any disk which is associated with dbvgextend for Node 2

Just start the cluster service for both the nodes :-

Node 1 = smitty clstart ( wait for sometime to complete)

Node 2 = smitty clstart ( wait for sometime to complete)

Thats all ..as per my understanding ..

Sam

 

[khalidaaa]

I'm afraid that you can't add hdisk14 and hdisk15 to the VG - dbgv on both nodes because hdisk14 and hdisk15 are different disks on both boxes!?! They have different pvids!

and the same for hdisk16 and 17!

Node 1:
root@Node01$ lspv
hdisk14 00012b3a85df3ad3 None
hdisk15 00012b3a85dfaf3a None
hdisk16 00012b3a874f1ecd None
hdisk17 00012b3a874f69f1 None


Node 2:
root@webaix02$ lspv
hdisk14 00012b3a874f1ecd None
hdisk15 00012b3a874f69f1 None
hdisk16 00012b3a85df3ad3 None
hdisk17 00012b3a85dfaf3a None

you should do the following on node 1:
extendvg dbvg hdisk14
extendvg dbvg hdisk15
extendvg dbvgextend hdisk16
extendvg dbvgextend hdisk17

but on node 2 you should do this:
extendvg dbvgextend hdisk14
extendvg dbvgextend hdisk15
extendvg dbvg hdisk16
extendvg dbvg hdisk17

otherwise, you will have to rmdev -d for each disk and try to arrange them if you want them to be the same! if you don't mind the confusion then they will still work fine!

I did it on my cluster without stopping the hacmp!

I accidently faced this problem after doing this:

http://www.tek-tips.com/viewthread.cfm?qid=1311133&page=2

which they said its a known bug on hacmp 5.3 but i don't think that you would face this! any way to fix this problem i had to stop and start the cluster so in either way just try the above first and if you face any problem you can stop and start the cluster! But it should work without doing so! I know sometimes it is difficult to bring down the cluster as this means downtime for the users! (This is not the case thought for hacmp 5.4 where you can leave the resource in an unmanaged state!)

Regards,
Khalid

 

[samirm]

Khalid,

In my case since the SSA disks are getting accessed by both the servers (Node -1 and Node -2) I didn't add those disk information for node 2. Only I did at Node-1.
I did cluster stop and did exort - import VG at Node -2 and it took the disk information. And started the cluster at both the node.

I am holding 4.5 HACMP version. I didn't try that when HACMP is running.

Khalid : One more thing, do you have any handy document, just to configure the HACMP for both the nodes from Scratch ? We are going to do this for two P550 servers, in which we will make 2 LPARS in each node.

Node -A : -
LPAR 1A = will be with Production database
LPAR 2A = Will be with test database.

Node -B :-
LPAR 1B = Will be with production Application.
LPAR 2B = Will be with test application.

We are going to implement HACMP 5.3 in between LPAR1A and LPAR1B.

Any idea ?

Sam

 

[khalidaaa]

Sam,

I'm using HACMP 5.3, so this might work a little bit smother than the 4.5 when it comes to adding disks! any way, regarding you request, I would be happy to assist, though it would be a very very long document but i will try my best

I would suggest you first try what i'm gonna say on the 2A and 2B LPARs (as they are the test ones

0) first install HACMP 5.3 on both machines.

1) Make sure that 2A and 2B have the same disks shown on both systems using the lspv

eg:

lspv

hdisk1 00012b3a85df3ad3 None
hdisk2 00012b3a85dfaf3a None

Note: make sure that both have the same PVIDs on both nodes

2) get one unused major number on both nodes using lvlstmajor command (eg let's say 45)

3) On 2A create an enhanced concurrent volume group (say HAvg)
smitty vg --> select Add an Original vg -->
vg name: HAvg
partition size:
pv names: hdisk1 hdisk2
Activate vg Auto: no
vg major number:45
Create vg concurrent capable? enhanced concurrent

4) vary on the vg and create a jfslog and format it using logform

5) create a test lv and fs for that lv (remember to set the mount auto. at system restart to No)

6) mount the fs and then unmount it and then varyoff the HAvg

7) Now go to Node 2B and issue the command
importvg -V 45 -y HAvg hdisk1

8) varyonvg HAvg
chvg -an HAvg
mount /fs (not needed just to make sure every thing is ok)
umount /fs
varyoffvg HAvg

oh no i gotta go now sorry but just do this far and let me know how it goes

I will be back to finish the rest

Regards,
Khalid

 

[khalidaaa]

oh yeah

here you go

http://www.tek-tips.com/faqs.cfm?fid=5514

regards,
Khalid

 

[mag007]

I usually use CSPOC to do anything related to LVM.

hth

 

[samirm]

Khalidaaa,

I really appreciate your patience to write this section for me. If you have any good pdf or link ( URL) let me know. I will also review the document which you sent. This activity will happen in Jan 2007 mid - last week. The servers are on the way to arrive in my locations.

Meanwhile if you have any good document / URL would be great.

Thanks again for your support.

Sam

 

[djml]

Thanks for everybody for the help.

I wish to everybody a great new year.