Dynamic Alert Rules

[changwuf31]

So we can allow something like

IF node_load1 > 2 * count(node_cpu{mode="idle"}) by (instance) FOR 5m

[brian-brazil]

That's a valid rule, though load average isn't a particularly useful metric for alerting these days. We'd generally encourage alerting on something that directly affects users such as latency or error rates.

[changwuf31]

It runs without failure, yes.. but it does not return the expected result.
And yes, we'd love to use latency and error rates as alert trigger, still trying to figure how best to implement that.

[brian-brazil]

Try:

node_load1 > on (instance) 2 * count by (instance)(node_cpu{mode="idle"})

[changwuf31]

@brian-brazil It works, thank you 👍

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.