Resource alert metric reference
The Metric type field on the Resource alert create page depends on the Type you select. The tables below list the metrics for each resource type, and in the UI each metric name carries its unit in parentheses, for example CPU usage (%).
Virtual Machine
Apply To: All instances or individual Virtual Machines.
| Metric | Unit | Description |
|---|---|---|
| CPU Usage | % | CPU utilization |
| RAM Usage | % | RAM utilization |
| Network Bandwidth In | Mbps | Inbound bandwidth across NICs |
| Network Bandwidth Out | Mbps | Outbound bandwidth across NICs |
| Total Network Bandwidth | Mbps | Total inbound and outbound bandwidth across NICs |
| Total packet In/Out per second | packet/s | Total inbound or outbound packets per second |
| Drop packet In/Out per second | packet/s | Inbound or outbound packets dropped per second |
| Disk IOPS | IOPS | Disk read/write operations per second |
| Disk Read Bandwidth | MBps | Disk read bandwidth |
| Disk Write Bandwidth | MBps | Disk write bandwidth |
Load Balancer
Apply To: All load balancers or individual Load Balancers.
| Metric | Unit | Description |
|---|---|---|
| HTTP Request | rps | Requests per second to the Load Balancer |
| HTTP code status (1xx) | rps | 1xx responses per second |
| HTTP code status (2xx) | rps | 2xx responses per second |
| HTTP code status (3xx) | rps | 3xx responses per second |
| HTTP code status (4xx) | rps | 4xx responses per second |
| HTTP code status (5xx) | rps | 5xx responses per second |
| Active connection | connection | Open connections to the Load Balancer. One connection can carry many requests |
| Percentage of backend down | % | Percentage of backends that are down (unhealthy) |
| Number of backends down | backend | Number of backends that are down (unhealthy) |
| Total packet In/Out per second | packet/s | Total packets the Load Balancer handles In or Out, useful for spotting unusual load |
| Drop packet In/Out per second | packet/s | Packets dropped per second, useful for spotting connection disruption or overload |
| Provisioning status Pending | N/A | Alerts when Provisioning status is Pending, meaning the Load Balancer is being created, updated, or reconfigured |
| Operating status Unhealthy | N/A | Alerts when Operating status is Unhealthy |
Enable Provisioning status Pending and Operating status Unhealthy first. They tell you the Load Balancer has stopped working, not just that load is high.
For Active connection, Total packet In/Out per second, and Drop packet In/Out per second, the alert adds an Action field (None, Auto-resize) that upgrades the Load Balancer package automatically when the metric crosses the threshold. See Create a Resource alert.
Object Storage
Apply To: the Object Storage region, for example HN-02.
| Metric | Unit | Description |
|---|---|---|
| Object Storage Usage | % | Percentage of Object Storage capacity used |
| Object Storage Usage | GB | Object Storage capacity used |
Object Storage alerts accept only Project-level recipients.
VPN Site-to-Site
| Metric | Unit | Apply To | Description |
|---|---|---|---|
| CPU usage | % | VPN Gateway (auto-filled) | CPU utilization of the VPN Gateway |
| RAM usage | % | VPN Gateway (auto-filled) | RAM utilization of the VPN Gateway |
| Packet loss | packet/s | VPN Gateway (auto-filled) | Packets lost per second between the two VPN endpoints |
| VPN Gateway status Offline | N/A | VPN connection | Alerts when the VPN Gateway goes Offline |
| VPN Connection status phase 1 Offline | N/A | VPN connection | Alerts when Phase 1 (IKE) of the VPN connection is Offline |
| VPN Connection status phase 2 Offline | N/A | VPN connection | Alerts when all Phase 2 (IPsec) tunnels are Offline. If at least one tunnel is Online, no alert is sent |
Auto Scaling
Apply To: Auto Scaling groups in the VPC.
| Metric | Unit | Description |
|---|---|---|
| CPU | % | Average CPU utilization across all instances in the Auto Scaling group |
| RAM | % | Average RAM utilization across all instances in the Auto Scaling group |
Auto Scaling alerts add an Action field (None, Scale In, Scale Out) that adjusts the number of instances automatically when the metric crosses the threshold. See Create a Resource alert.
See also
- Create a Resource alert — steps to create an alert and what each condition field means.