> The entire point is to alert when things are crashing, so you would want it to be highly available. As in, running on more than one node. An important questi

> The entire point is to alert when things are crashing, so you would want it to be highly available. As in, running on more than one node. An important question to ask is how much availability are you actually gaining from the setup. It wouldn't be the first time I see a system moving from single-node to multinode and being less available than before due to the extra complexity and moving pieces.