Observe
Metrics
CPU, memory, network and disk charts for applications and databases, and the Dashboard's four tiles.
Updated
Every application and database is measured from the moment it exists, with nothing to switch on. Its resource use appears in two places: the service's own Metrics section, and the Resource use panel on the Dashboard, which draws the same charts for any one service you pick.
A compose stack is not measured. It runs several containers, and one chart would mix their numbers, so read a stack through its logs instead.
The four charts
| Chart | What it draws | Unit |
|---|---|---|
| CPU | Processor time the container uses | Percent of one core, so 100 is one full core and 200 is two |
| Memory | Memory in use, against the total the container reports | Megabytes; the label reads "MB used of" followed by that total |
| Network | Data received (In) and sent (Out) | Megabytes per second |
| Block I/O | Data read from (Read) and written to (Write) disk | Megabytes per second |
The platform samples each container every 30 seconds. Network and disk figures arrive as running totals since the container started, so Cloud draws the change from one sample to the next, divided by the seconds between them. When a container restarts its totals start again from zero, and the chart shows that step as no traffic rather than a negative rate.
Time range
The Time range menu picks how far back the charts reach:
| Time range | Samples drawn |
|---|---|
| Last 30 minutes | 60 |
| Last 90 minutes | 180 |
| Last 6 hours | 720 |
Six hours is as far back as the charts go. They show the samples that existed when the page loaded; pick the range again or reload the page to draw newer ones.
When there is nothing to chart
The panel says why in one sentence and draws nothing else:
| Sentence | What it means |
|---|---|
| Metrics are not available on this platform yet. | Metrics collection is not configured on the platform, so no service has charts. |
| The agent has nothing for this service yet. Points appear a minute or two after it starts. | Cloud adds a new application or database to collection within a minute of its creation, and the first samples follow. On a compose stack this sentence never goes away, because stacks are not measured. |
Any other sentence is Cloud's answer when it could not read the samples, such as "Your workspace is still being prepared."; the page itself keeps working.
Cloud sends no alert when a metric crosses a threshold, and it runs no uptime checks. Notifications cover deployments, domains and backups.
The Dashboard
The Dashboard opens on four tiles under Overview, each counting the organization the console is in, which the line under its title names:
| Tile | What it counts |
|---|---|
| Projects | The organization's projects |
| Services running | Services whose state is Running; the line under the count gives your total, such as "of 4" |
| Deployments | Deployments requested in the last 24 hours |
| Failed | Deployments requested in the last seven days that ended Failed |
Under them, Resource use holds the four charts for one service. It opens on your first service; the Service menu ("Pick a service") lists every service as its project, a slash, its name and its state, and the Time range menu works as above. The line under the panel, "Usage hours and spend arrive with billing.", predates billing: hours and spend are on Settings, then Billing, described in Billing.
The Storage usage card below shows the disk used by persistent data, volume by volume; Volumes explains it. The Dashboard then lists your most recently updated projects and your latest deployments.