hdisk mismatch on 2 nodes

[newcg]

Hi,
I have 2 AIX nodes running HACMP. How do I get the 2 nodes with the same hdisk#.

node1: active. with all FS mounted.
hdisk0 00c018cacacbabf3 rootvg active
hdisk1 00c0116ad056e074 datavg1 concurrent
hdisk2 00c011aeda31a029 datavg2 concurrent
hdisk3 00c0121eca9b0291 datavg3 concurrent

node2:
hdisk0 00c0143d2a3b902a rootvg active
hdisk1 00c0116ad056e074 datavg1 concurrent
hdisk2 00c0121eca9b0291 datavg3 concurrent
hdisk3 00c011aeda31a029 datavg2 concurrent

On Node 2, hdisk2 and hdisk3 are switched compared to NODE1. How do I get NODE2 to look like NODE1 when I do a "lspv".
Thanks.

 

[khalidaaa]

first of all i would say that this configuration will cause no problems at all when using HACMP. Its just inconvenient for you i believe!

You can do this by removing the disks from the ODM and rediscover them again!

you will have first to down the cluster then use rmdev (I beleive you will have to use this rmdev –R –dl darX (X is according to your dar setup))

and then cfgmgr to rediscover them.

Good Luck

Regards,
Khalid

 

[newcg]

I'm not sure about it'll be a problem for the hacmp. I tried to create a new lv on datavg2(20G free avail on it) on the active node1, and it wouldn't let me. I only need to create 10G lv/fs.

But if I create any size lv in datavg1 (20 G free space avail) I have no problem.
So, I'm thinking the mismatch may cause it.

If I down the cluster on node2 and only do:
rmdev -dl hdisk2
rmdev -dl hdisk3

and reboot, will it work..? I don't know what the darX is for. I do a "lsdev" and I don't see any dar.
thanks.

 

[newcg]

Khalid,
I did a rmdev -Rdl hdisk2 and hdisk3 on node2 and when I do a cfgmgr, it goes back to the same hdisk. No change. any ideas?

 

[khalidaaa]

Having the disks in that order has nothing to do with the problem u are facing! That's for sure!

I assumed that you have a SAN connecting the two nodes of the cluster and SAN uses these terms dar and dac.

you have to remove the parent of hdisk2 and hdisk3 to be able to do sort this out

what is the output of this

lsdev -C -F parent -l hdisk2
lsdev -C -F parent -l hdisk3

when you find the parents from the output above use the command

rmdev -Rdl hdisk2_parent_name

then run cfgmgr and let me know the result!

But again i'm pretty sure that this is not the cause of the problem of not being able to create the lv! Are you creating the lv using the CSPOC?

regards,
Khalid

 

[mag007]

I agree with Khalid. HACMP + AIX only care about PVId, not the disk name.

 

[newcg]

Thanks khalid. I had to rm the hdisks on the active node also. I rmdev -Rdl on all the shared hdisks on both servers and do a cfgmgr and when it came up, they both looked the same. And I was able to create "lv" and "fs" on the "vg" I wasn't able to last time. I think it was due to the mismatch. Now, I;m able to create lv/fs, etc on the hdisks that "matched" without any problem using cspoc or the normal way.
Thanks.

 

[khalidaaa]

I think i had this problem once! I was unable to create an lv on on of my resource group's vg. But all i had to do is to down the cluster and start it again for this to take effect! I'm again sure that it wasn't because of the disk mismatch! but in away, removing those disks might set up the VGDA on both nodes! I guess this happens when you try to create an lv manually!

An advice! Always Use CSPOC for your LVM. It is much safer!

Regards,
Khalid