Dear
peter_grundler_automicThank you for the response.
I fully understand the typical 2 node scenario where the one node is active and the other is in active-standby(NWP). The document is written by assuming that scenario is what guess and due to the fact ,Unfortunately we cannot figured-out how exactly this feature help in providing high-availability in slightly complex scenarios.
You know,
everyone preferred on distributing workloads
on multiple of small servers than on few big fat servers today due to various benefits ( mostly on cloud environments ). Deploy to multiple availability zones with minimum( or one) node to take initial switch over workload and then provision more nodes as per job processing demand ( may be as per the job schedule timing etc) .
So , I'm trying to understand when it come to more than two nodes scenario .
4 nodes AE system. 2 node in Active and 2 on active-standby(NWP)
mode scenario . Basically 2 AE node licenses and 2 NWP license.
My assumption is that,One of the Active node(very first node), one WP will assume as pWP (
usually the very first WP started in the AE system and connected to the DB).
and all the remaining WP(s) on two active nodes will then be active and do the job processing as per its role. (However as per the document only one node WP(s) can be active).
and all the remaining WP(s) on two active-standby(NWP ) nodes will be in NWP(s) mode andwill notdo any job processing. and they are waiting for active node(s) to fail to be assume as active node. Document says "If the primary computer that includes the active WPs fails, the nonstop processes take over their role" . However it does not describe what it means by "primary computer fails", is it that pWP of the node fails. or all WP(s) fails or all CP(s) and WP(s) fails . or the node not reachable from the NWP node over IP ( but node may be running still). or the node is completely down etc..
Rgds,
IP