Adding a disk to a MetroCluster VG

[marius etsebeth_1]

Hi people,

How do I add a disk to a VG in an active Metrocluster (CA/XP)? I currently have 22 disks assigned to my device group, but they have all been utilized by various VG's.

I want to add the disk to the VG and create 2 new LV's, but I need to know the steps for the MetroCluster portion and is it necessary to take the cluster down for it?

Any help would be appreciated...

[Ivan Krastev]

Do you have Raid Manager running for data replication management ?

regards,
ivan

[marius etsebeth_1]

Ivan,

Apologies. The answer is "yes".

[melvyn burnard]

well first step is you have to have the pair created, so that the CA is copying across fpr the Pvol to the Svol.

Then you need to do your pvcreate, and vgextend, to add in the disk on the Pvol side.
You also need to add it into to your horcm?.conf file, remembering to do it on all nodes.
You then need to export using -p this VG, and import it on all the nodes on the same site (i.e. still at the Pvol side.)
you would then need to import it onto the nodes in the Svol site, either by doing a pairswap first, or doing a pair suspend with a -rw option, so that LVM can read/write to the PV's in the VG.
Only then can it be used by the package.

And yes I guess the package that uses this VG must halted to make certain of these changces

[marius etsebeth_1]

Melvyn,

Thanks for the reply so far, but I have more questions. Please bear with me as I'm learning here.

I'm attaching my horcm.conf file and the output of raidscan in a single document.

A snippet from the raidscan is seen below:

PORT# /ALPA/C,TID#, LU#.Num(LDEV#....)...P/S, Status,Fence, LDEV#,P-Seq#,P-LDEV#
CL1-P /e1/ 4, 1, 5.1(1536)..........SMPL ---- ------ -----, ----- -----
CL1-P /e1/ 4, 1, 6.1(1674)..........SMPL ---- ------ -----, ----- -----
CL1-P /e1/ 4, 1, 7.1(1605)..........P-VOL PAIR DATA 1605, 10243 3738
CL1-P /e1/ 4, 2, 0.1(1743)..........P-VOL PAIR DATA 1743, 10243 3346

1.) How do I translate

CL1-P /e1/ 4, 1, 5.1(1536)..........SMPL ---- ------ -----, ----- -----

into a physical volume (/dev/rdsk/cxtyd0)?

2.) Let's say I have identified a free PV on my Primary and secondary hosts each.

Host A
------
/dev/rdsk/c17t6d4 e1 06 34 CL1P 10:0c OPEN-V 00010215

and

Host B
------
/dev/rdsk/c13t3d6 c9 03 1e CL2P 0e:54 OPEN-V 00010243


So, in my horcm.conf file I would add for e.g. (refer the the horcm.conf):


p01oms p01oms_025 10215 10:0c

and

p01oms p01oms_025 10243 0e:54

Now, to re-read the config, I have to restart raid manager with horcmshutdown.sh .

Would this affect the package or cluster i.e. take it down? This is important to know, as it is a PROD environment.
Sadly, we do not have a MetroCluster in a DEV environment (due to costs implications).

From there I presume the following steps (dev_group=p01oms ; dev_name=p01oms_025):

1.) paircreate -g p01oms p01oms_025 -vl -c 15
2.) pvcreate ..
3.) vgextend ...
4.) vgexport -p...
5.) pairsplit -g p01oms -rw

On secondary node:

6.) vgimport -s -m
7.) vgchange -a y
8.) vgcfgbackup
9.) vgchange -a n
10.) pairresync -g p01oms -c 15

done - hopefully

[melvyn burnard]

You should have xpinfo on these nodes, use that to help translate or provide a relationship.

[marius etsebeth_1]

Melvyn,

Yes, I do have xpinfo, but I needed "raidscan -fx" to get the LDEV's in HEX, which I missed (a bit embarrased at this point)

Now, does the "horcmshutdown.sh " affect the package in any way or is it safe to stop / start the RaidManager instance to re-read the config?

[melvyn burnard · Accepted solution]

Now, does the "horcmshutdown.sh " affect the package in any way or is it safe to stop / start the RaidManager instance to re-read the config?

No, it does not affect the packages in any way UNLESS during the shutdown and restart time a failure were to occur AND a package/site switch were required.
So you have around a 3-6 second window. But as you have to do this on each node, it will only affect anything on that node, so a very limited possibility of an issue here.

[Ivan Krastev]

Whan using paircreate use only for the new devices with -d option:

â d[s] pair_volSpecifies a paired volume name written in the configuration definition file. The command executes only for the specified paired volume.

As far as i remember last time no the downtime at all was required - nor the package or cluster.

regards,
ivan

[marius etsebeth_1]

Hi all,

OK, with Melvyn and Ivan's help and after a lot of reading, I managed to do it.

I'll award points as appropriate.

Attached is a step by step guide for the next noobie :)

[marius etsebeth_1]

Hi all,

OK, with Melvyn and Ivan's help and after a lot of reading, I managed to do it.

I'll award points as appropriate.

Attached is a step by step guide for the next noobie :)