Oracle ASM and RAW devices.

[tom quach_1]

Hi All,

Please advice how i can use my raw devices for the oracle ASM.
i have HP Storageworks Virtual Array 7110.
when i created a LUN for 10Gb for the system. it generated 2 devices
/dev/rdsk/c6t0d2
/dev/rdsk/c4t0d2

for Oracle ASM, i just need to change ownership and add to Oracle ASM. But which one?
If someone has done this before.
Please advice.

Thanks in advance.
Regards,
Tom

[Sandman!]

They two device files that you are seeing look like Alternate Paths to the same LUN. Do you have two HBAs on your HPUX server? Just add one of them paths to the Oracle ASM software and you should be fine. Please post the output of the following command:

# ioscan -funC ext_bus

[A. Clay Stephenson]

These are PROBABLY devices nodes which correspond to the same physical LUN. It is common for arrays to create multiple paths to the same LUN so that if one path fails an alternate is available. I say probably because I didn't carry out the operation but your array management software should be able to tell you. Another method (if you know these these are indeed unused LUN's) is to use dd to copy data using one of the device nodes and use dd to read it back using the other device node. If the data match, you are talking to the same physical LUN.

[tom quach_1]

Thanks Sandman and A. Clay.

there are 2 HBA on the host.
yes i could pick any device for the ASM and it would work. But i worry about fail over when one path goes dead.
yes, these 2 devices are from the same physical LUN.
i guess the question is how to make it failover when ASM only know of 1 device. we have EVA3000 with securePath and it taking care of this by combine the 2 devices into one. is there a similar way to do it.

#ioscan -funC ext_bus
Class I H/W Path Driver S/W State H/W Type Description
=========================================================================
ext_bus 0 0/0/1/0 c8xx CLAIMED INTERFACE SCSI C896 Ultra Wid
e LVD
ext_bus 1 0/0/1/1 c8xx CLAIMED INTERFACE SCSI C896 Ultra Wid
e Single-Ended
ext_bus 2 0/0/2/0 c8xx CLAIMED INTERFACE SCSI C87x Ultra Wid
e Single-Ended
ext_bus 3 0/0/2/1 c8xx CLAIMED INTERFACE SCSI C87x Ultra Wid
e Single-Ended
ext_bus 4 0/10/0/0.8.0.110.0 fcparray CLAIMED INTERFACE FCP Array Int
erface
ext_bus 5 0/10/0/0.8.0.255.6 fcpdev CLAIMED INTERFACE FCP Device In
terface
ext_bus 6 0/12/0/0.8.0.108.0 fcparray CLAIMED INTERFACE FCP Array Int
erface
ext_bus 7 0/12/0/0.8.0.255.6 fcpdev CLAIMED INTERFACE FCP Device In
terface


and these are 2 path from 2 HBA with the same physical LUN

disk 5 0/10/0/0.8.0.110.0.0.2 sdisk CLAIMED DEVICE HP A6189B
/dev/dsk/c4t0d2 /dev/rdsk/c4t0d2

disk 6 0/12/0/0.8.0.108.0.0.2 sdisk CLAIMED DEVICE HP A6189B
/dev/dsk/c6t0d2 /dev/rdsk/c6t0d2



Thanks,
Tom

[Sandman!]

Using LVM should provide you with the fault tolerance you need in case the primary path goes down. You can read more about it in this thread:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=692210

~hope it helps

[tom quach_1]

Thanks Sandman.

if i use LVM, pvlink will be perfectly fine. In my case, i use Raw devices in Oracle ASM. i could use one link for the ASM and it would work fine also, but in case this link dies there is no failover for the survive link.
how to combine the 2 path to become one as securepath on my EVA3000 does?

Thanks,
Tom

[Sandman!]

You are right Tom...don't know what I was thinking. Somehow got the lvols confused with the raw device special files. Imho you need to install the Auto Path software. It is the Secure Path equivalent on VA 7000 class of storage systems. Read more about it in the HP whitepaper below:

http://h20000.www2.hp.com/bc/docs/support/SupportManual/lpg28858/lpg28858.pdf?jumpid=reg_R1002_USEN

~cheers

[tom quach_1]

Hello,

now i got autopath working but i still have to devices.
please advice how i can map it so it present me with one device.
thanks in advance.
Tom,


#autopath display -l 6F00-0200-0000-0000-1D00-0000-0000-0000

==================================================================
HPswsp Version : A.3.0F.02F.02F
==================================================================
Array WWN : 060B-0000-151A-6F50

==================================================================
Lun WWN : 6F00-0200-0000-0000-1D00-0000-0000-0000
Virtual Device File : /hpap/dsk/hpap0
Load Balancing Policy : Round Robin
Lun Timeout : Infinite Retry (-1)
==================================================================
Device Path Status
==================================================================
/dev/dsk/c4t0d2 Active
/dev/dsk/c6t0d2 Active

[Duncan Edmonstone]

Tom,

You already have a single device:

Virtual Device File : /hpap/dsk/hpap0

You just need to change you ASM_DISKSTRING parameter in your ASM instance to point to /hpap/dsk

I guess you may need to chnage the permissions/ownership on that device as well

HTH

Duncan

[Yogeeraj_1]

Hi Tom,

Now that you have identified the Virtual Device File : /hpap/dsk/hpap0

The Oracle data directory could be setup the following way:
$ ln -s /hpap/dsk/hpap0 /u02/oradata/asmdisks/disk00
$ ln -s /hpap/dsk/hpap1 /u02/oradata/asmdisks/disk01
$ ln -s /hpap/dsk/hpap2 /u02/oradata/asmdisks/disk02
$ ln -s /hpap/dsk/hpap3 /u02/oradata/asmdisks/disk03

And the following command needs to be executed after each reboot:
# chown oracle.dba /u02/oradata/asmdisks/*

You need to ensure that the ownership of block devices is changed to oracle:dba or oracle:oinstall. Otherwise Oracle will not be able to access the block devices and ASM disk discovery will not list them. (NB. In linux, we had to ensure that the ownership of block devices was set after each reboot since Linux changes the ownership of block devices back to "brw-rw---- 1 root disk" at boot time. You will need to verify if the behaviour is the same on HP-UX).

and finally, to create your ASM disk group, you will need to set the Disk Discovery Path to "/u02/oradata/asmdisks/*"

Hope this helps too!

kind regards
yogeeraj

[tom quach_1]

Thank you all for the info.

question:
1-
would you want to use character device for the ASM disk group
/hpap/rdsk/hpap0 and not the block device /hpap/dsk/hpap0 ?
2-
Would anyone use this device /hpap/dsk/hpap0
to create volume group?

Thanks,
Tom

[Sandman!]

>question:
>1-
>would you want to use character device for the ASM disk group
>/hpap/rdsk/hpap0 and not the block device /hpap/dsk/hpap0 ?
>2-
>Would anyone use this device /hpap/dsk/hpap0
>to create volume group?

For the first question it depends on the application usage of data. Typically if the application reuses data then it is best to go with the block device file since data is buffered and cache hits would mitigate performance penalties. The rules for using raw over block device files and vice-versa remain the same.

No you would not use the virtual device file presented by AutoPath to create the VGs because this is storage side multi-pathing while LVM is host side multi-pathing software. You would create the VGs using LUNs presented by ioscan and vgextend the alternate links.

~hope it helps

[Duncan Edmonstone]

Tom,

oops you're right - you would use the raw device special file for ASM, not the block.

HTH

Duncan

[tom quach_1]

Thank you Sandman and Duncan,

One more question on securepath
from securepath below, i have 2 path active/active
but when create vg group with these 2 active devices. It became active/alternate Link
why not active/active on vg group.

Thanks,
Tom



#autopath display

==================================================================
HPswsp Version : A.3.0F.02F.02F
==================================================================
Auto Discover : OFF
==================================================================
Array Type : VA
Array WWN : 060B-0000-151A-6F50
Path Verification Period : 00:10
==================================================================
Lun WWN : 6F00-0200-0000-0000-1D00-0000-0000-0000
Virtual Device File : /hpap/dsk/hpap0
Load Balancing Policy : Shortest Service Time
Lun Timeout : Infinite Retry (-1)
==================================================================
Device Path Status
==================================================================
/dev/dsk/c4t0d2 Active
/dev/dsk/c6t0d2 Active


--- Physical volumes ---
PV Name /dev/dsk/c4t0d2
PV Name /dev/dsk/c6t0d2 Alternate Link
PV Status available
Total PE 2559
Free PE 2309
Autoswitch On

[Duncan Edmonstone · Accepted solution]

Tom,

Secure Path and LVM are separate products with no rel knowledge of each other.

When you created the VG, you *could* have just added on of the didk devive special files, and Secure Path would still have managed failover to the other.

Of course another admin would need to know about Secure Path and run the commands to spot that the other device special file is pointing to the same actual LUN, and there's no harm in adding the alternate link as well. Secure Path should still be the component that manages failover to the other path as its further down the stack, but in theory LVM could as well - it's kind of a 'belt and braces' (is that 'belt and suspenders' in US english???) approach.

HTH

Duncan

[Sandman!]

The feature that you are referring to is concerned with static or dynamic load balancing. The difference is that LVM is an example of an active/passive while Secure Path is an example of an active/active storage management system. LVM only provides static load balancing multiple paths while Secure Path provides dynamic load balancing over multiple paths (storage-side controllers). You can read more about it in the whitepaper below:

http://docs.hp.com/en/11822/storagemulti-pathing-HPUX-SG.pdf

[tom quach_1]

Thank you very much for your helps!
Regards,
Tom