Node shutdown and quorum loss
In this example, the quorum node (node2)
is shutdown, causing quorum loss on node1.
Running the same example on node2, the output
is similar to:
_failed_ _n_ 199.18.1.2 _nn_ node1 _rc_ 668 _t_ 1121974459 _tu_ 616867
If the -p flag is not specified, the
output is similar to:
mmpmon node 199.18.1.2 name node1: failure status 668 timestamp 1121974459/616867
lost quorum
In this scenario there can be a window where node2 is down and node1 has not yet lost quorum. When quorum loss occurs, the mmpmon command does not attempt to communicate with any nodes in the node list. The goal with failure handling is to accurately maintain the node list across node failures, so that when nodes come back up they again contribute to the aggregated responses.