questions about the vpar

[Matthew_50]

Hello, gurus,

few vpar questions as following.

1. In vpar, can locate vpar’s root vg ( 2 disks mirror ) ? if I would like to have two vpar on an npar, and each vpar with self mirrored vg00, Which means, 6 disks are require ? ( npar0 needs 2, vpar0 needs 2, and vpar1 needs 2 )
2. When vparcreate, support 2 bootpath ( primary, alternative ? )
3. Base on 2 vpar in each npar, for example, npar0 with vpar0, vpar1, npar1 with vpar2, vpar3. Is it possible that vpar0 and vpar2 config as cluster, vpar1, and vpar3 config as cluster ? any sense ?
4. Continue with question3, the vpar with service guard failover mechanism.
4.1 config mcsg between npar0, and npar1, when npar0 hardware failure, failover npar0’s vpar setting ( vpar0, vpar1) to npar1 ?
4.2 any other mcsg config possibility?

Thanks, and plz let me know if any confuses :)

[Torsten.]

Hi,

0) a OS instance running on a vPar is just the same like an OS installation without vPar from the OS point of view.

1) you need separate disks on separate SCSI controllers or on the SAN for each Vpar.

2) yes - see (0)

3) Its possible, but makes no sense, because if your whole server fails, your cluster will be down

4) see (3)

[Rick Madison]

Matthew,

In your example in #1, you will only need (4) disks not (6). You'll install an OS on the nPar0, create your vPars, then convert nPar0 to a vPar, becoming vPar0. You convert it by running a `mkboot -a "hpux /stand/vpmon -a" /dev/rdsk/c#t#d#`.

[Raj D.]

Mathew,

After installing the vpar on the npar0 you will have two vpar on that npar0 and the npar0 will be itself a vpar , so you will need 4 disk. I agree with Rick. and you need to use the mkboot option ( mkboot -a "hpux /stand/vpmon -a" /dev/rdsk/c#t#d#`).
Also you can verify with #vparstatus -v command and lifcp /dev/dsk/cXtXdX:AUTO -



Cheers,
Raj.

[siva0123]

Mathew,

Regarding the MC service guard configuration as you have suggested , it is possible but what is the use?

Service guard is for high availability , and if you are going to configure a cluster in a single hardware where is the high availability if someone blows up ur server to pieces!!!:-((( ( Not necessarily a blow .. just think of a power failur for the complete server)

always it is advisible clusters are configured on different servers and if posiible on different locations.

There are clusters which are even spanning across continents!!!