Disk Enclosures
1748269 Members
3674 Online
108760 Solutions
New Discussion юеВ

MSA30 U320 redundant RAID controller/dual domain active/passive failover configuration?

 
SOLVED
Go to solution
Daniel Pocock
Occasional Advisor

MSA30 U320 redundant RAID controller/dual domain active/passive failover configuration?



The MSA30 quickspecs refer to `controller duplexing'. Is this similar to the dual domain capability of the StorageWorks 60 SAS array? How can it be used?

Can someone describe whether it is possible to have an active/passive configuration like the following:
- two HP Proliant servers (e.g. DL585)
- each server has a Smart Array controller (e.g. 6402)
- both controllers/servers on the same SCSI bus
- one controller is active, the other passive (not doing anything)
- if the server with the active controller fails, the passive controller can be instructed to become active and provide continued access to the array
- the operating system is Linux

There are some specific questions about this configuration:
- Does the dual bus (MSA30 DB) allow both controllers to address all drives? This seems contrary to the meaning of dual bus, but then I saw the words `controller duplexing'.
- If using the single bus (MSA30 SB) what type of cable or connector would be required (HP part number?) to join both controllers to the array?
- How do I manually configure each array controller to use a distinct SCSI ID?
- Is there some way that the Smart Array 6402 controller can be told to be active or passive? Can this be done from the BIOS and the HP Array Configuration Utility in Linux?

I've seen descriptions of the dual domain and dual port features of the SAS arrays and drives, but a lot of those products don't appear to be in the market yet. Meanwhile, we have an existing investment in MSA30 SB (single bus) arrays and want to configure them for higher availability, without waiting another 6 months for the SAS products to be put on the market.

http://h18000.www1.hp.com/products/quickspecs/11738_div/11738_div.html

Dual Bus I/O Module

The dual bus I/O module allows two SCSI channels to access a single MSA30. This allows for greater fault tolerance (controller duplexing) or provides access from two different servers to two separate storage areas.
3 REPLIES 3
Khairy
Esteemed Contributor
Solution

Re: MSA30 U320 redundant RAID controller/dual domain active/passive failover configuration?

hi,

HP has another msa30 called MSA30MI (multi-initiator) for this kind of setup (cluster/shared scsi storage).

You can't use msa30 db/sb for shared storage.

You can also upgrade you msa30 sb to multi initiator. Replace SB module to a MI module.

Refer attach doc.

It intend to be cheap/simple setup for Integrity server clusters. But i dont find any support docs about using in proliant....

To do this, you have to set scsi ID connect to msa at each server to different so that it wont conflict with each other. Set ID 6 to 1st server and ID 7 to 2nd server scsi adapter.

And yes, sa6402 raid adapter is supported for msa30Mi.

Rgds



Daniel Pocock
Occasional Advisor

Re: MSA30 U320 redundant RAID controller/dual domain active/passive failover configuration?



I found the MSA30 MI quickspecs. It says that Smart Array (hardware RAID) controllers are not supported. It also seems to be missing Linux on the list of supported OSes.

Has anyone configured this device successfully with Linux?

I'm not aware of any technical reason why a multi-initiator configuration shouldn't work with a Proliant server and Linux, I suspect these exclusions are intended to encourage sales of HP's high end servers.

http://h18004.www1.hp.com/products/quickspecs/11587_div/11587_div.html

NOTE: Smart Array controllers (A9890A, A9891A, A7143A, A9826A, A9825A) are NOT supported in HP Serviceguard/Multi-initiator configurations
Daniel Pocock
Occasional Advisor

Re: MSA30 U320 redundant RAID controller/dual domain active/passive failover configuration?



I've also found this support matrix on HP's web site.

It suggests that even with HP-UX, you can't have RAID and MI configuration together.

I suspect it is related to the working of the battery powered write cache on the RAID cards - the server that fails (e.g. server A) may have left the drives in an inconsistent state, and when server A restarts, it may try and flush the cache data out to the array, corrupting data that was written by the server B.

http://www.hp.com/products1/serverconnectivity/support_matrices.html