Use the gce_instance monitored resource type

cloudfoundry-community / stackdriver-tools

Stackdriver Nozzle for Cloud Foundry Loggregator, Host Monitoring Agents BOSH Release

Apache License 2.0

21 stars 13 forks source link

Use the gce_instance monitored resource type #63

Closed evandbrown closed 6 years ago

evandbrown commented 7 years ago

Using gce_instance instance as the monitored resource type (rather than the global monitored resource) increases throughput as nozzle instances are added since the Stackdriver API shards based on the gce_instance's instance_id label.

johnsonj commented 6 years ago

This has been addressed where it makes sense. The nozzle now records its own metrics sharding on gce_instance. The metrics coming out of Cloud Foundry need to stay global so that the calculations aren't messed up.

This was more relevant before we had robust filtering, grouping of metrics. Now that the nozzle fits in the default metric quota we've greatly reduced the load that inspired this issue.