Install the HP Lefthand P4000/P4500 Failover Manager to prepare for disaster

Imagine 4 P4500 Lefthand’s spread over 2 locations and operating in 1 management group and 1 cluster with Network RAID-10 (2-Way Mirror) Data Protection Level. How many managers do you need to keep the SAN operational in case of a disaster (destruction of 1 site)?

The answer is 3. If you install a manager on every machine, you’re still 1 manager short in case of the disaster. It’s possible to use a virtual manager, but the SAN still goes down at the time of the disaster. The virtual manager can only help to start it again.

The logical thing to do here, is to install a Failover Manager. Put this manager in a third location and you’re protected against a disaster in any of those 3 locations.

For our pleasure, HP created a Virtual Appliance to be installed on a virtual platform. Download it here. In my case I used a simple HP ML110 G5 with 2x Intel Xeon 3065 CPU’s, 8GB RAM, 2x 500GB SATA disks and 2 GBit NICs with VMware vSphere Hypervisor (Free ESXi) as a dedicated server for the Virtual Appliance. And even that is overkill. You can easily make it work with 1 dual core CPU and half the RAM.

Setting up the complete environment and integrating it into the existed infrastructure took me less then a day. Configuring the Virtual Appliance is not a lot more then assigning an IP address.

Advertisements

About Yuri de Jager
Technology Addict

10 Responses to Install the HP Lefthand P4000/P4500 Failover Manager to prepare for disaster

  1. Heino says:

    Hi,

    How long did it take for your FOM to join the management group ?

    BR
    Heino

  2. Heino says:

    Thanks for the quick answer.

    Then i have a problem.
    after 1 hour it is still saying “Waiting for 1 manager to join”

    Do you know by any change what can be wrong ?

    BR
    Heino

    • Yuri de Jager says:

      I can’t say for sure. But I would definitely check your versions and go through all settings again.

    • DS says:

      Resurrecting an old comment, but most likely it’s a network issue (the management interface cannot be contacted by the FOM during setup. Make sure they’re on the same vlan or a vlan that can route to the storage server management interfaces.

  3. Nick says:

    Hello Yuri,

    we are interested in the HP P4300 G2 starter system because of the hardware redundancy in first place. As far as I’ve understand, one system is a set of two servers which build up a cluster of two nodes. Do I also need a FOM running in this case to have the hardware redundancy working between the two nodes? Example: one Server goes down because of a mainboard failure, the second server comes in place automatically and takes over all services.

    We’re using Open-e DSS for this at the moment. It works fine but we need something bigger, so we use the chance to have a look what else could be suitable for us.

    TIA+BR
    Nick

    • Yuri de Jager says:

      Hello Nick,

      Yes, you do. I take it you use 2-way replication. In matter of fact, VMware has published a document about this specifically: http://communities.vmware.com/docs/DOC-13342

      Pay extra attention the connection to your SAN, set the datastore’s Path Seleciton to Round Robin.

      • Nick says:

        Hello Yuri,

        thank you for your answer. Unfortunately the FOM procedure would add to much overhead to us, just to achieve a simple 2 node redundant iSCSI Storage. It seems to me the P4300 solution is a bit overdrawn for what we need. So I think we will keep searching a bit for an other solution or just stay with Open-e on bigger servers.

        Best regards
        Nick

  4. Mario says:

    Hi Yuri,

    we do have the exact same setup (4x P4500 in 1 Cluster, 1x Management Group, 2x Locations) and we have some problems understanding all the types of “managers” in CMC. There are “normal” Managers, virtual Managers and failover Managers. What do I really need? And what’s the difference between them?
    We don’t wan’t to install an additional FOM and HP told us that it is not necessary. We could just stop the normal Manager on the second P4500 in the second site and this should be sufficent for building the Quorum.
    Why do I need the managers?
    We have 2 scenarios to consider:
    1. crash of 1x P4500
    2. destruction of a whole site (2x P4500)

    In case 1 the Network Raid 1+0 would save our data. Do I need any managers here?
    In case 2 also the Network Raid should save our data.

    What are the managers for? Just for deciding when a “Split-Brain” happens?

    I’d really appreciate your answer…

    Regards
    Mario

    • Yuri de Jager says:

      Hello Mario,

      I am by no means a Lefthand expert, so I would defenitely recommend to discuss your concerns with you supplier.

      Remember to configure your nodes in the CMC to the correct site. If you put eg. node 1 and 3 in site 1 and node 2 and 4 in site 2, and configure this in the CMC accordingly, Lefthand will configure Network RAID-10 2-Way Mirror for you correctly. Writes to a node in site 1 will also be written to a node in site 2. This way, you will be protected against scenario 1.

      For scenario 2, you need a FOM if you want your SAN to stay up. The FOM needs to be on a seperate, 3rd site. Preferrably physically, but it can be a logical site too. This way, there will always be 3 managers to keep the SAN up in case of scenario 2, as I explained in my post.

      In the case where the FOM is only on another logical network (so still in site 1 or 2), you have a 50% change that the SAN will not go down in scenario 2.

      You can, of course, go without FOM. But then it is recommended to install (not activate!) a Virtual Manager. In case of scenario 2, your SAN will go down, but by starting the Virtual Manager, you will have created the neccesary 3 managers to keep the quorum and you can start the SAN again.

      I think it’s pretty obvious why I would recommend a FOM over a Virtual Manager =)

      Good luck!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: