I was quite surprised the first time when the metric karpenter_interruption_received_messages{message_type="SpotInterruptionKind"} didn't match karpenter_nodeclaims_terminated{reason="interruption"}.
Looking through the eventbridge configuration, it's quite clear why the karpenter_interruption_received_messages metric is aggregated at the account level. Perhaps the documentation could be updated to make it clear?
Description
How can the docs be improved?
I was quite surprised the first time when the metric
karpenter_interruption_received_messages{message_type="SpotInterruptionKind"}
didn't matchkarpenter_nodeclaims_terminated{reason="interruption"}
.Looking through the eventbridge configuration, it's quite clear why the
karpenter_interruption_received_messages
metric is aggregated at the account level. Perhaps the documentation could be updated to make it clear?https://github.com/aws/karpenter-provider-aws/blob/e326781350cf113a2769860588c994d1a02612f5/website/content/en/docs/reference/metrics.md?plain=1#L101-L102
Alternative would be to filter out non-Karpenter-related events from the
karpenter_interruption_received_messages
metric.