How to define the same special file disks on 2 servers !!

[Ivan Azuara]

Hi !!

I need to cover a requeriment about the configuration of Oracle RAC on hpux 11.31. I need to have the same special files disks on each hpux server that form the cluster. This zone of disks will be managed by the ASM.

This is my scenario:

Node A (hpux 11.31)

/dev/rdisk/disk974 /dev/rdsk/c31t0d3
/dev/rdsk/c24t0d3
/dev/rdisk/disk975 /dev/rdsk/c31t0d4
/dev/rdsk/c24t0d4
/dev/rdisk/disk976 /dev/rdsk/c31t0d5
/dev/rdsk/c24t0d5
/dev/rdisk/disk977 /dev/rdsk/c31t0d6
/dev/rdsk/c24t0d6
/dev/rdisk/disk978 /dev/rdsk/c31t0d7
/dev/rdsk/c24t0d7
/dev/rdisk/disk979 /dev/rdsk/c31t1d0
/dev/rdsk/c24t1d0
/dev/rdisk/disk980 /dev/rdsk/c31t1d1
/dev/rdsk/c24t1d1

Note: The c24* disks are the alternate path.

Node B (hpux 11.31)

/dev/rdisk/disk60 /dev/rdsk/c18t0d3
/dev/rdsk/c14t0d3
/dev/rdisk/disk61 /dev/rdsk/c18t0d4
/dev/rdsk/c14t0d4
/dev/rdisk/disk62 /dev/rdsk/c18t0d5
/dev/rdsk/c14t0d5
/dev/rdisk/disk63 /dev/rdsk/c18t0d6
/dev/rdsk/c14t0d6
/dev/rdisk/disk64 /dev/rdsk/c18t0d7
/dev/rdsk/c14t0d7
/dev/rdisk/disk65 /dev/rdsk/c18t1d0
/dev/rdsk/c14t1d0
/dev/rdisk/disk66 /dev/rdsk/c18t1d1
/dev/rdsk/c14t1d1

Note: The c14* disks are the alternate path.

My external storage is a USPV from Hitachi, and my storage administrator makes the necesary configuration to guarantee that each LUN from the storage can be access by the 2 nodes. But the ASM configuration only it can define one name by each disk for the 2 nodes that will be handled by the ASM.

Can you help me with this issue ?

Thank's in advance !!

[Torsten.]

A possible solution:

http://www.google.de/url?sa=t&source=web&cd=6&ved=0CE0QFjAF&url=http%3A%2F%2Fh20000.www2.hp.com%2Fbc%2Fdocs%2Fsupport%2FSupportManual%2Fc02273756%2Fc02273756.pdf


vgcdsf - a LVM tool to convert persistent DSFs in a volume group to their corresponding cluster DSFs

[Jeff_Traigle · Accepted solution]

Firstly, on 11.31, I'd recommend using the agile device files instead of the legacy ones so you take advantage of the built-in multipathing. These are located in /dev/rdisk.

Once you've identified which LUNs are associated with which device file on each node, you can create an additional device file with a common name and change ownership and permissions on them.

Example:

On node 1, /dev/rdisk/disk170 is LUN 1.

Identify major and minor number of this device file.

# ls -l /dev/rdisk/disk170
crw-r----- 1 bin sys 13 0x00002a Sep 24 17:48 disk170

Create new device file. Choose a naming convention that works for you. We find it useful to identify them as oracle devices, type of device (vote, data, flashback), incremental identifier, and array the LUN is on.

# mknod /dev/rdisk/ora_data_disk01_array1 c 13 0x00002a
# chown oracle:dba /dev/rdisk/ora_data_disk01_array1
# chmod 660 /dev/rdisk/ora_data_disk01_array1

On node 2, /dev/rdisk/disk270 is LUN 1.

Identify major and minor number of this device file.

# ls -l /dev/rdisk/disk270
crw-r----- 1 bin sys 13 0x000030 Sep 24 17:50 /dev/rdisk/disk270

Create new device file.

# mknod /dev/rdisk/ora_data_disk01_array1 c 13 0x000030
# chown oracle:dba /dev/rdisk/ora_data_disk01_array1

The DBAs now have the common device file /dev/rdisk/ora_data_disk01_array1 to use in their RAC configurations.

[Ivan Azuara]

Hi Jeff

Thank´s for the advice. With this procedure i can guarantee that the alternate paths works correctly because i´ll use the agile format that handled the multipath, correct ??.

Best Regards !!

[Manix]

read about CDFS .It`s a new feature i saw in MCSG 11.20

Cluster-wide device special files (cDSFs) are persistent device special files applied
across a set of nodes. That is, they ensure that the same piece of storage has the same
devicefile name on all of the nodes that share it; no matter how many paths there are
to the device, the same cluster DSF is used to address it. If the device is moved, the
same cDSF still addresses it.
Create the cDSFs.

NOTE: cDSFs apply only to shared storage;
â ¢ If the cluster does not exist yet, specify the name of each prospective node,
for example:
cmsetdsfgroup -n node1 -n node2 -n node3 -n node4
â ¢ If the cluster does exist, you can simply run:
cmsetdsfgroup -c

Do read release notes for MCSG 11.20 edition
for details & implementations.

http://h30499.www3.hp.com/t5/System-Administration/dev-cdisk-disk/m-p/4755089#M388582

[Jeff_Traigle]

That's correct. With the agile device files, there's no need to worry about managing the alternate paths (PV links) because the multipathing capability of the mass storage stack of the OS handles it for you. One of my favorite features of HP-UX 11.31.

BTW, I just noticed I missed the chmod command in my example for node 2. It'd be the same as on node 1, of course.

[Ivan Azuara]

Hi Manix

Unfornately i don't have the Service Guard product.

Regards.

[Pete Randall]

They're just arbitrary names - rename them! You can call them /dev/george and /dev/bill and /dev/ivan or anything else your heart desires.


Pete

[Ivan Azuara]

That's right!!. I can use any name for the devices. But i think that the procedure of Jeff adds a level of security considering that i have to grant permissions to the dba administrator over the special disk devices.

Regards !!

[Ivan Azuara]

Jeff

Finally i don't need to consider the /dev/disk directory ??. This is because the RAC is going to work only with the raw devices, correct ??.

Regards !!