Building a Highly Available Failover Cluster Solution With WS2012 From The Ground Up

Some notes taken from TechEd NA 2012 WSV324:

I won’t blog too much from this session.  I’ve more than covered a lot of it in the recent months.

Cluster Validation Improvements

• Faster storage validation
• Includes Hyper-V cluster validation tests
• Granular control to validate a specific LUN
• Verification of CSV requirements
• Replicated hardware aware for multi-site clusters

CSV Improvements

• No external authentication dependencies for improved performance and resiliency
• Multi-subnet support (multi-site clusters)

Asymmetric Cluster

BitLocker on CSV

This will get the BitLocker status of the CSV:

manage-bde –status C:ClusterStorageVolume1

This will enable BitLocker on a CSV:

manage-bde –on C:ClusterStorageVolume1 –RecoverPassword

You get a warning if you try to run this with the CSV online.  You need the volume to be offline (Turn On Maintenance Mode under More Actions when you right-click the CSV) … so plan this in advance.  Otherwise be ready to do lots of Storage Live Migration or have VM downtime. 

NOTE! A recovery password is created for you.  Make sure you record this safely in a place independent from the cluster that is secure and reliable.

Get the status again to check the progress.

It’s critically important that you add the security descriptor for the cluster so that the cluster can use the now encrypted CSV.  Get that by:

get-cluster

Say that returns the name HV-Cluster1.

Now run the following, and note the $ at the end of the security descriptor (indicating computer account for the cluster):

manage-bde C:ClusterStorageVolume1 –protectors –add –sid HV-Cluster1$

That can be done while the CSV is encrypting.  Once encrypted, you can take it out of maintenance mode.

AD Integration

• You now can intelligently place Cluster Name Objects (CNO) and Virtual Computer Objects (VCO) in desired OUs. 
• AD-less Cluster Bootstrapping allows you to run/start a cluster with no physical domain controllers.  This gets a justifiable applause It’s great news for branch offices and SMEs.
• Repair action to automatically recreate VCOs
• Improved logging and diagnostics
• RODC support fro DMZ and branch office deployments

Node Vote Weight

• In a stretch or mult-site cluster, you can configure which nodes have votes in determining quorum.
• Configurable with 1 or 0 votes.  All nodes have a vote by default.  Does not apply in Disk Only quorum model.
• In the multi-site cluster model, this allows the primary site to have the majority of votes.

Dynamic Quorum

• It is now the default quorum choice in WS2012 Failover Clustering
• Works in all quorum models except Disk Only Quorum.
• Quorum changes dynamically based on nodes in active membership
• Numbers of votes required for quorum changes as nodes go inactive
• Allows the cluster to stay operations with >50% node count failure

Thoughts:

• I guess it is probably useful for extremely condensed cluster dynamic power optimisation (VMM 2012)
• Also should enable cluster to reconfigure itself when there are node failures

Configuration:

EnableDynamicQuorum edit a cluster common property to enable dynamic quorum

DynamicWeight Node private property to view a node’s current vote weight

Cluster Scheduled Tasks

3 types:

• Cluster wide: On all nodes
• Any node: On a random node
• Resource specific: On the node that owns the resource

PowerShell:

• Register-ClusteredScheduleTask
• Unregister-ClusteredShceduledTask
• Set-ClusteredScheduledTask
• Get-ClusteredScheduledTask