[nagar_yash]

We have two IBM AIX machines(lets call A and B) with 4 LPARs on
each. These LPARs are clustered using HACMP to provide the high
availability.

Each resource groups has serviceiplabels , unique application and
shared VGs.


.
Problem description:
A Machine- has UX1, UX6, UX7, UX3 as the name of LPARs.
B Machine- has UX2, UX5, UX8, UX9 as the name of LPARs.


A Machine and SAN env B Machine and SAN env

UX1 clustered to failover----> UX5

UX6 <----clustered to failover UX2

One resource group is almost 95% in disk usage(VGs are mirrored
across A and B, SAN storage to provide high availability) in
between UX1 and UX5 and going to *die* at any time.

We have identified 4 disks mirrored across UX2 and UX6 which are
unused(there are no LV and/or data) and can be re-assigned to UX1
and UX5.

Please help me to take all precuationary measure to move the LUNs from clustered UX2-UX6, to UX1-UX5.

Thanks
-Yashpal

[ross.mather]

First most important step in doing this: Backup everything you have - and check it.

Now what type of storage are you using? DS4000, DS8000? Something else?

Is there Virtual I/O involved?

The important thing to know is which LUNs you are moving around. Make sure that the donating cluster is completely working before adding disks to the receiving cluster.

Take your time, plan your actions and your back-outs if things go wrong. What happens if the wrong LUN is deleted from the LPAR?

[nagar_yash]

Hi!

Thanks for your reply.

Storage is fastT700 on both the P650- AIX 5.2. Both the machines can see both the storage boxes.

4 disks which we have identified to be moved are not part of any resource groups, backup is also not the problem, VIO is not involved.

We have checked fget_config -vA and gathered the LUNs of the free disks.

I have planned like

• Take the snapshot of existing config files, ensure last backup of rootvg went Ok
• Removing the volume groups
• Remove the PVID entries from ODM from UX2 & UX6
• Rescan and/or cfgmgr subsystem to check if the disks has appeared well on UX1 & UX5
• Move the disks into a the volume group

is this Ok or i have missed something?

Regards
Yashpal

[cdelgadop]

Hi

Sorry i don't understand what you want.

Do you mean that those 4 disks (SAN Based ?? i mean, are those LUNs ??) assigned to the Volume Groups where UX2 and UX6 resides (UX2 and UX6 are in the same VG ??) are free ??.

If so, and depending on the version of HACMP, you can use the smitty cspoc to perform some LVM tasks online on the VG structure, so you can delete those 4 disks form the VG and assign to the VG where UX1 and UX5 are defined. Then using the same cspoc options increase the size of the Logical Volumes and File Systems with problems.

I don't understand what you mean by " UX1 clustered to failover----> UX5" ??

Please explain.

Hope this helps

[ross.mather]

Makes sense. I'd make sure you back up everything as the most likely error in the whole thing is that the wrong LUN will be removed by the SAN admin.

[nagar_yash]

Sorry for the confusion.

Yes, the 4 disks are SAN based LUNs which are currently allocated to UX2-UX6 cluster and not part of any resource group (RG).

UX2-UX6 has 2 VGs, with those 4 disks, and there is no LV and data.

UX2/UX6 reports these disks as:

UX2 > fget_config -vA
---dar0---
User array name = 'A_FAStT700'
dac0 ACTIVE dac2 ACTIVE

Disk DAC LUN Logical Drive
utm 31
hdisk33 dac0 44 UX2_projecti_2_m
hdisk34 dac0 45 UX2_projectt_2_m
hdisk35 dac2 46 UX2_projectd_5_m
hdisk36 dac0 47 UX2_projectx_m
---dar1---
User array name = 'B_FAStT700'
dac1 ACTIVE dac3 ACTIVE

Disk DAC LUN Logical Drive
utm 31
hdisk58 dac1 44 UX2_projecti_2
hdisk59 dac3 45 UX2_projectt_2
hdisk60 dac1 46 UX2_projectd_5
hdisk61 dac3 47 UX2_projectx

tubairux6 > fget_config -vA
---dar0---
User array name = 'A_FAStT700'
dac0 ACTIVE dac2 ACTIVE

Disk DAC LUN Logical Drive
utm 31
hdisk44 dac0 44 UX2_projecti_2_m
hdisk45 dac0 45 UX2_projectt_2_m
hdisk46 dac2 46 UX2_projectd_5_m
hdisk47 dac0 47 UX2_projectx_m

---dar1---
User array name = 'B_FAStT700'
dac1 ACTIVE dac3 ACTIVE

Disk DAC LUN Logical Drive
utm 31
hdisk70 dac1 44 UX2_projecti_2
hdisk71 dac3 45 UX2_projectt_2
hdisk73 dac3 47 UX2_projectx
hdisk72 dac1 46 UX2_projectd_5

The cluster is in the form of two sites A & B.

site A site B

UX1 ---clusterd--- UX2 (cluster where space is required)
UX6 ---clusterd--- UX5 ( 4 Free disks are here)
UX7 ---clusterd--- UX8
UX3 ---clusterd--- UX9

AIX 5.2, HACMP 5.3



Thanks
Yashpal

[nagar_yash]

My apologies:

Last portion should be read as:


site A site B

UX1 ---clusterd--- UX5 (cluster where space is required)
UX6 ---clusterd--- UX2 ( 4 Free disks are here)
UX7 ---clusterd--- UX8
UX3 ---clusterd--- UX9

[Adam-Smith]

Hi!
My name is Adam!