Hard Failover Cannot Be Safely Performed Despite Allowing Unsafe

Couchbase Server
1

Background
I want to autoscale query nodes for my cluster.

The cluster is 2 nodes at maximum and one node at minimum.

I’m using "implementationVersion": "6.5.2-6634-enterprise".

Issue

If the target node is up, then it will be added and removed from the cluster according to a schedule.

But there are cases when my provider may eliminate the query machine when it’s a member of a cluster.

In the case I want to perform a hard failover.

But, when I send a request to controller/failOver with allowUnsafe=true, I get the response Cannot safely perform a failover at the moment. Yet this is why I specified allowUnsafe=true.

I see at Hard Failover | Couchbase Docs that “In the event of a cluster’s Master Services not being able to contact a majority of the cluster’s nodes (sometimes referred to as a quorum failure ), an attempted failover will not be executed.”

Is this what I’m experiencing?

What can I do to perform an unsafe failover for a query node in these conditions:

  1. One permanent node.
  2. One unresponsive query node.

Thank you.

2

I was sending this payload with Python Requests library:

{
   'otpNode': target_otp_node,
   'allowUnsafe': True
}

I changed True to a string 'true' and a forcible rebalance initiated:

{
   'otpNode': target_otp_node,
   'allowUnsafe': 'true'
}