Monitoring node failures

As mentioned previously, Marvel keeps track of nodes even after they leave the cluster. This is useful when working with large clusters that have many nodes to keep track of.

We will demonstrate how the Marvel Nodes dashboard displays node failures by shutting down elasticsearch-node-01:

# From elasticsearch-node-01
sudo service elasticsearch stop

The Nodes dashboard in Marvel now looks like this:

Here we can see Marvel indicates elasticsearch-node-01 used to be part of the cluster, but is currently offline.

Elasticsearch-head, on the other hand, shows us the cluster in a yellow state, but does not indicate that elasticsearch-node-01 was ever part of the cluster:

Note

Elasticsearch-head only displays elasticsearch-node-02 and elasticsearch-node-03, not elasticsearch-node-01.