Good afternoon,
Our company has implemented distributed between the DC database cluster (with uplinks between local nodes in each of the DC)
The solution to all comfortable with except one thing:
When network problems occur at the providers - if they are long-lasting (such as a physical on / off on the uplink equipment, block uplinks, etc)
Bootstrap node begins to fall apart (even if a quorum in terms of where there is a node - dialed (2 + 2 nodes arbitrator)
Total in the cluster at this point 7 nodes
I would be grateful if any share methods of solving this problem