Mackerel supports obtaining and monitoring Compute Engine metrics with Google Cloud Integration. When integrating with Google Cloud Integration, billable targets are determined using the conversion 1 Instance = 1 Standard Host. Google Cloud Integration can also be used together with mackerel-agent. In such cases, system metrics and custom metrics are integrated into one host (with no redundant charges).
Please refer to the following page for Google Cloud Integration configuration methods and a list of supported services.
Google Cloud Integration
Required API
The following API must be enabled in order to integrate Compute Engine.
- Compute Engine API(compute.googleapis.com)
Required role
The following role is required for the service account in order to integrate Compute Engine.
- Compute Viewer(roles/compute.viewer)
Obtaining metrics
The metrics obtainable with Google Cloud Integration’s Compute Engine support are as follows. For Metric explanations, refer to the Google Cloud documentation.
The maximum number of obtainable metrics is 10 + 11 × [disk name] + [packet drop source]
| Graph name | Index type | Metric name in Mackerel | Unit |
|---|---|---|---|
| CPU | instance/cpu/utilization | gce.instance.cpu.used | percentage |
| Disk bytes | instance/disk/read_bytes_count instance/disk/write_bytes_count |
gce.instance.disk_io.[disk name].read gce.instance.disk_io.[disk name].write |
bytes |
| Disk ops count | instance/disk/read_ops_count instance/disk/write_ops_count |
gce.instance.disk_ops.[disk name].read gce.instance.disk_ops.[disk name].write |
integer |
| Disk throttled bytes | instance/disk/throttled_read_bytes_count instance/disk/throttled_write_bytes_count |
gce.instance.disk-throttled_io.[disk name].[throttling source].read gce.instance.disk-throttled_io.[disk name].[throttling origin].write |
bytes |
| Disk throttled ops | instance/disk/throttled_read_ops_count instance/disk/throttled_write_ops_count |
gce.instance.disk-throttled_ops.[disk name].[throttling source].read gce.instance.disk-throttled_ops.[disk name].[throttling origin].write |
integer |
| Network bytes | instance/network/received_bytes_count instance/network/sent_bytes_count |
gce.instance.network.received gce.instance.network.sent |
bytes |
| Network packets count | instance/network/received_packets_count instance/network/sent_packets_count |
gce.instance.network_packets.received gce.instance.network_packets.sent |
integer |
| Uptime | instance/uptime | gce.instance.uptime.uptime | integer |
| Firewall dropped bytes | firewall/dropped_bytes_count | gce.firewall.dropped | bytes |
| Firewall dropped packets count | firewall/dropped_packets_count | gce.firewall_packets.dropped | integer |
| Mirroring bytes | mirroring/mirrored_bytes_count | gce.mirroring.mirroed | bytes |
| Mirroring packets | mirroring/mirrored_packets_count | gce.mirroring_packets.mirrored | integer |
| Mirroring packets dropped | mirroring/dropped_packets_count | gce.mirroring_packets_dropped.[packet drop source] | integer |
- Enter the disk device name linked to the Compute Engine instance in place of [disk name] in "Metric name in Mackerel".
- Enter an identifier for the packet drop source in place of [packet drop source] in "Metric name in Mackerel".
- Enter an identifier for the throttling source such as
PER_GBorPER_VMin place of [throttling source] in "Metric name in Mackerel".
