Failure notifications in multi-node clusters

An article comprising at least one non-transitory machine-readable storage medium comprising instructions executable by a processor of a node of a multi-node cluster to:

• determine if an incoming node failure notification has been received, wherein the incoming node failure notification indicates that another node of the multi-node cluster is in failure;
• based on the determination that the incoming node failure notification has been received, freeze execution of all processes on the processor and flush a cache on the node to a non-volatile memory;
• determine if the node is in failure;
• and
• based on a determination that the node is in failure,
• generate an outgoing node failure notification, and
• send the outgoing node failure notification to each node in the multi-node cluster.

A method of a node of a multi-node cluster, the method comprising:

• determining if an incoming node failure notification has been received from another node of the multi-node cluster, wherein the incoming node failure notification indicates that the another node of the multi-node cluster is in failure;
• determining if a system failure notification has been received from a cluster management interface, wherein the system failure notification indicates that a system critical process alert has been triggered;
• determining if the node is in failure;
• based on the determination that the node is in failure, generating an outgoing node failure notification and sending the outgoing node failure notification to each node in the multi-node cluster;
• and
• based on one of the determination that the node is in failure, the determination that an incoming node failure notification has been received, and the determination that a system failure notification has been received,
• freezing execution of all processes on a processor on the node, and
• flushing a cache on the node to a non-volatile memory.