How does a pri­mary and / or sec­ondary get selected?

• The first 5 hosts that join the VMware HA clus­ter are auto­mat­i­cally selected as “pri­mary nodes
• All the oth­ers are auto­mat­i­cally selected as “sec­ondary nodes”
• When you do a recon­fig­ure for HA the pri­mary nodes and sec­ondary nodes are selected again, this is random

What’s up with these pri­maries and secondaries?

• Pri­mary nodes hold clus­ter set­tings and all node states which are synced between primaries
• Sec­ondary nodes send their state info(resource occu­pa­tion) to the pri­mary nodes
• Nodes send heart­beats to each other, pri­mary nodes send heart­beats to pri­mary nodes only and sec­ondary also only to pri­mary. And they do this every sec­ond. (Which is a change­able value: das.failuredetectioninterval)

So what if a pri­mary node fails, will a sec­ondary be promoted?

• No, there will only be a new pri­mary appointed when the failed one is removed from the clus­ter. A sec­ondary will be pro­moted to pri­mary at random.

But what if all my pri­mary nodes fail?

• This is an unad­dressed issue, that’s the rea­son why you can only account for 4 host fail­ures within a clus­ter! There needs to be at least one primary!

So when does the gate­way come in play?

• Actu­ally the gate­way, which is the default “iso­la­tion address”, will only be used when an iso­la­tion has occurred. So when the AAM client thinks it’s iso­lated it will check the iso­la­tion addresses.

From — Yel­low Brick — a very lovely VMWare ori­ented site.