First make sure you are familiar with the design of alert-manager by reading the DESIGN doc.
Items below here should include target links from alerts.
Every 15s the alert-to-pubsub instance should receive a POST request from the local Prometheus instance and then convert all of those alerts into PubSub events, including a healthz PubSub event. This alert is fired if a location, like skia-corp has failed to generate a healthz event recently. Check that both Prometheus and alert-to-pubsub are running in the designated cluster. Also check the PubSub Topic and Subscriptions.