HPE EVA Storage

Redundancy Configuration

 
CraigScott
Occasional Contributor

Redundancy Configuration

Hi All,

I have a similar post on the Blade Forum however they are 2 slightly different questions so hope its ok...

Wonder if you can help - We currently have the following equipment:

C7000 Enclosure
1x Brocade SAN Switch
1x MSA2312fc with 1 Controller
BL460c G6 Blades with 1 QLOGIC QMH2562 Card in Each.

The switch currently has zoning disabled.

Im aware that we are running a risk of only having 1 SAN switch and 1 controller in the SAN, however, is there a way of connecting up this equipment to allow for a fibre cable failure.

At the moment we have Port A1 and A2 going from the SAN to the switch. The Luns have been created and are mapped to the blades via both ports.

When I go into the Qlogic configuration on the blades I can see both Ports on the SAN and both Luns presented in each.

Within windows, the drives are duplicated. I am presuming this is because its being presented with the same drive twice, once from each port on the SAN.

If I unplug the fibre cable from port A1 on the SAN, will it auto switch to start using port A2?

How does it work if we were to purchase another controller for the SAN? Will we still see the devices presented twice? Im unsure on how I should configure the switch with Zones etc?

Kind Regards,

Craig

6 REPLIES 6
Uwe Zessin
Honored Contributor

Re: Redundancy Configuration

Craig,

if you are seeing two disks and have only provided one MSA volume - the server is missing the multipath filter - Windows alone will NOT automatically do path failovers. This component will care about redundant paths and present one device to the file system.

If you buy another controller, present the volumes correctly and miss to install the filter, yes, you will see 4 disks.

I've done a quick look at HP's web site and it looks like you need to select the correct MSA2000 DSM if you are running Windows 2003:
http://h20000.www2.hp.com/bizsupport/TechSupport/DriverDownload.jsp?prodNameId=421495〈=en&cc=us&prodTypeId=18964&prodSeriesId=421492&taskId=135

I have not tried it myself, but I guess you also can install the "HP MPIO DSM (Device Specific Module) Manager for Full Featured DSMs" so that you have a management plugin for the MMC.

The integration for Windows 2008 should be described in one of the MSA manuals.


My zoning rules are:

- single-initiator / single-target
-- a port from a blade server is an initiator
-- a port from the MSA is a target

- create zones using the port WWNs (World Wide Names)
-- that way you can move devices between switch ports in the same fabric (e.g. move a blade to a different slot if you suspect a connectivity problem or move the MSA to different switch ports) without needing to adapt the zoning
.
CraigScott
Occasional Contributor

Re: Redundancy Configuration

Hi Uwe,

Thank you very much for your reply.

So im seeing the disk twice, thats normal at least!

I have looked at the MPIO software and will give it a go when im back in work tomorrow. I was under the impression that failover would happen automatically and it would be controlled by the Qlogic cards etc.

Do I need to have MPIO installed on every server I wish to use failover with - regardless if I have 1 or 2 Brocades / MSA Controllers ?

Are you aware if there is any MPIO software for Hyper V Server (Version that installs just with a command prompt)?

Will my current setup work?:

1 MSA controller, using both A1 and A2 ports connected to 1 brocade switch - Will this be successful to prevent a cable failure? I understand that if the controller or brocade goes im in deep water but im wondering if my configuration will technically work.

I will check out the zoning tomorrow and have another go at it - If I currently have zoning turned off, I thought I could still move switch ports / blades around without encountering problems?

Kind Regards,

Craig
Sajeev2007
Frequent Advisor

Re: Redundancy Configuration

Craig,

you will need to use some multipathing software as UWE said.

Microsoft has one ( free with the OS )

The other one is Symantec/Veritas Enterprise Administrator

- Sajeev
Steven Clementi
Honored Contributor

Re: Redundancy Configuration

I concur, as usual...

Other comments:

"I have looked at the MPIO software and will give it a go when im back in work tomorrow. I was under the impression that failover would happen automatically and it would be controlled by the Qlogic cards etc."

Failover will happen automatically, but it is the multipath driver/filter/module that handles the actual failover.

"Do I need to have MPIO installed on every server I wish to use failover with - regardless if I have 1 or 2 Brocades / MSA Controllers ?"

If you have 2 paths to your storage now with only 1 switch and 1 controller, than yes. You need MPIO on every server. It's free, so no worries.

"Are you aware if there is any MPIO software for Hyper V Server (Version that installs just with a command prompt)?"

The same package "should" work with Hyper-V, though I have limited experience with it.

"Will my current setup work?:

1 MSA controller, using both A1 and A2 ports connected to 1 brocade switch - Will this be successful to prevent a cable failure? I understand that if the controller or brocade goes im in deep water but im wondering if my configuration will technically work."

Yes, to an extent. The cable from A1 is redundant with the cable from A2.


"I will check out the zoning tomorrow and have another go at it - If I currently have zoning turned off, I thought I could still move switch ports / blades around without encountering problems? "

If you are in a homogeneus environment (all windows or all linux or all sun or all hp-ux, etc), then generally speaking you should be "ok" temporarily while you sort things out. Get the zoning done before you hit production. ;o)


Steven
Steven Clementi
HP Master ASE, Storage, Servers, and Clustering
MCSE (NT 4.0, W2K, W2K3)
VCP (ESX2, Vi3, vSphere4, vSphere5, vSphere 6.x)
RHCE
NPP3 (Nutanix Platform Professional)
CraigScott
Occasional Contributor

Re: Redundancy Configuration

Thanks for your replies everyone, finally starting to see the bigger picture!

Just one more question if thats ok?

I am looking at the zoning. I have created two alias's one called "Blades" in which I have assigned the WWNs of the Blade HBA's and another called "SAN" - Which I have assigned the WWN of the SAN controllers.

I have then created a zone and added both the Alias's into it - Finally, I added that zone into the Zone Config I created.

Can someone explan what

single-initiator / single-target

is?

In the SAN switch documentation it mentions I can go into the Zone Config tab and there should be a "Device Accessibility" option where the "Initiator/Target Acessibility Matrix for Config-Device Selection".

It states I should then be able to set which devices are the initiators and which are the targets.

However I cant find any such option?

If I right click on a Device (The WWN) and select Device detail, i can see that the blades have been assigned as an "initator" under Host vs. Target and the SAN is assigned as a Target.

Are these settings automatically applied based on the device type?

Regards,

Craig
Uwe Zessin
Honored Contributor

Re: Redundancy Configuration

> single-initiator / single-target

Current situation, so a single fabric:

Create 4 alias names for the blades and the MSA ports:
B1_1, B2_1, MA_0, MA_1.

Create four zones:
B1_1__MA_0: B1_1; MA_0
B1_1__MA_1: B1_1; MA_1
B2_1__MA_0: B2_1; MA_0
B2_1__MA_1: B2_1; MA_1

Create a configuration and add the zones:
CFG1: B1_1__MA_0; B1_1__MA_1; B2_1_MA_0; B2_1__MA_1

---
I love non-ambiguous names, so if you would introduce the second fabric it would look like:
Aliases: B1_2, B2_2, MA_1, MB_1.

(If you had a second controller module you would put Port_0 from controller A+B into the first fabric and Port_1 into the second. This will require changes in the first fabric!)

Four more zones:
B1_2__MA_1: B1_2; MA_1
B1_2__MB_1: B1_2; MB_1
B2_2__MA_1: B2_2; MA_1
B2_2__MB_1: B2_2; MB_1

Create a configuration and add the zones:

CFG1: B1_2__MA_1; B1_2__MB_1; B2_2_MA_1; B2_2_MB_1

In case of multiple fabrics I use the same configuration name. This will prevent an accidental merge, but the use of unique alias and zone names makes it easier if it is desired at a later time.
.