RAC cluster implementation

[Mawuse]

Hi,

Can somebody help me on the following customer specification:

1- Configuration of two RAC clusters ( 2 nodes each ) for ERP
2- LVM configuration of raw volumes for clustered databases
3- HP SGeRAC 11.16 clusters implementation
4- Installation and configuration of Oracle 10g CRS + Oracle 10g RAC binaries + patch 10.1.0.4
5- Creation of two databases
6- Integration of One Oracle database into the SGeRAC clusters


Questions

1- I use to configure Oracle Database package with HP Service Guard for automatic switch over . Now this is the case where by Oracle Real Application Cluster has to be configured. I don' t really know whether I have to configure package or packages.

2- It seems I don't really understand the customer specification.

3- Is it possible to configure a shared virtual disk on an EVA 4000 for simultaneous access from cluster nodes, in such a way that both nodes can have the shared storage file system mounted at the same time ?


Please enlighten me. Your contributions and clarification are welcomed.



Regards,
Mawuse

[IT_2007]

For the question 3.

- Is it possible to configure a shared virtual disk on an EVA 4000 for simultaneous access from cluster nodes, in such a way that both nodes can have the shared storage file system mounted at the same time ?

>> Yes. You can do it. One server it becomes as "server" and another node it shows as "Client" but you can mount them on both nodes with "read/write" access. Make sure that you activate the volume group with "-S" option in the package control file on both nodes.

[Mawuse]

Hi,

Thank you very much for your reply. It helped me make a giant step forward, but I still need one more help :
How will the file system be mounted on both nodes for access ?

Regards,
Mawuse

[IT_2007]

you need to specify mount point in the package control file so that when you activate package, it would be mounted on one node as server.

Once it is activated, you can write up a script or manually activate on second node so that it would activate as "client" and mount the file systems.

But remember that you need to specify them as OPS volumes in cmclconf.ascii file before you compile the cluster so that cluster knows them as OPS to achieve above requirements.

[Mawuse]

Thanks very much.
The activation of the vg with vgchange -a s vgdb on the client is successfull, but whenever I try to mount it on the client, it gives Error: /dev/vgdb/lvdb is corrupted.
Even after fsck the mount gives error again.

Regards,
Mawuse

[IT_2007]

you need to deactivate it and re-activate with -S option. Tt is upper case letter.

-S sharable Control the sharability of volume groups in a
high availability cluster. sharable can have
one of the following values:

y Mark each specified volume group as
sharable. The high availability
software must be running;
otherwise, the volume group is not
marked. Needs to be done on one
node only.

n Remove the shared attribute from
the volume group. The high
availability software does not need
to be running.

The volume group must be deactivated with the
-a n option before a -S y|n option can be
executed.

[Mawuse]

Thanks for your reply .
But whenever I try to deactivate it, it says :

vgchange: Failed to notify clvm daemon about volume group deactivation. Device busy. Couldn't deactivate volume group vgdb.

Regards,
Berenger

[IT_2007]

check vgdisplay -v vgdb and see what is the status of it. If you haven't mounted any related logical volums then you can deactivate it.

vgchange -a n /dev/vgdb

vgchange -c y /dev/vgdb

vgchange -c y -S y /dev/vgdb
vgchange -a s /dev/vgdb

[Bill Brunnler]

In The RAC clusrer I have set we used RAW devices. The cooked files systems are local only your oracle data, control,redo files are in a package. Each database would have two packages each packes with an innstance to the database. The packages would not fail over. The volume groups would need to be shared.

If the database is in archive mode I would recomend cfs. It makes life a lot easier .

[Mawuse]

Each of your contributions has really enlightened.

Thanks very much.