Questions and answers about Monitoring

General questionsGeneral questions

Can I get logs of my operations with services?Can I get logs of my operations with services?

Yes, you can request log records about your resources from Yandex Cloud services. For more information, see Data requests.

How do I view service dashboards?How do I view service dashboards?

Service dashboards are created automatically after resources are created in Yandex Cloud. Add a resource and update the Monitoring homepage.

The old data in Monitoring charts has disappeared. Why?The old data in Monitoring charts has disappeared. Why?

Metrics for which no new values have been received for 30 days are automatically deleted from Monitoring. For Yandex Managed Service for Prometheus®, this period is 60 days; for Yandex Compute Cloud, 7 days. For more information, see Deleting expired metrics (TTL).

There is also a non-configurable decimation mechanism that decreases the amount of stored data.

In Prometheus, metrics are not delivered, and I cannot create a workspace. Why?In Prometheus, metrics are not delivered, and I cannot create a workspace. Why?

This may happen if none of your cloud resources are sending monitoring data. Start using a resource and wait for the metrics to show up in Monitoring.

NavigationNavigation

Where can I view a VM's RAM chart?Where can I view a VM's RAM chart?

The Yandex Compute Cloud service can't measure vRAM consumption inside the guest operating system because for the service, memory consumption by the virtual machine is always the same: the one that is allocated the moment it is started.

You can configure collecting custom metrics yourself via the API, and for Linux-compatible operating systems, you can do it by a system metrics delivery agent.

Where can I view a VM's disk space chart?Where can I view a VM's disk space chart?

Monitoring can't access guest operating systems, so there is no such metric.

You can configure collecting custom metrics yourself via the API, and for Linux-compatible operating systems, you can do it by a system metrics delivery agent.

Metric collection and exportMetric collection and export

How do I add a VM's RAM metric?How do I add a VM's RAM metric?

Yandex Compute Cloud cannot measure vRAM consumption inside the guest operating system because, for this service, memory consumption by the virtual machine is always the same: the one that is allocated the moment it is started.

You can configure collecting custom metrics yourself via the API. In Linux-compatible operating systems, you can do it using a system metrics delivery agent. The agent allows you to collect statistics on the use of most system resources: CPU, RAM, network, and disk.

How do I add a VM disk space metric?How do I add a VM disk space metric?

Monitoring cannot access guest operating systems, which is why there is no such metric.

You can configure collecting custom metrics yourself Writing custom metrics via the API. In Linux-compatible operating systems, you can do it using a system metrics delivery agent. The agent allows you to collect statistics on the use of most system resources: CPU, RAM, network, and disk.

How do I collect custom metrics from my app?How do I collect custom metrics from my app?

Yandex Unified Agent supports metric collection in the Prometheus format and metric conversion to the Monitoring format. That means you can use Unified Agent to collect metrics from any application that provides them in the Prometheus format.

Use a client library for the programming language your application is written in and deliver metrics to Monitoring in the Prometheus format.

See examples in the following documentation sections:

• Delivering custom application metrics.
• Delivering metrics from HAProxy and other third-party applications.

Can I deliver Yandex Managed Service for Kubernetes cluster metrics to Monitoring?Can I deliver Yandex Managed Service for Kubernetes cluster metrics to Monitoring?

Yes, you can. To do this, use Metrics Provider. To install it, follow this guide.

How do I deliver third-party app metrics to Monitoring?How do I deliver third-party app metrics to Monitoring?

To deliver metrics from third-party applications to Monitoring, use Unified Agent with the metrics_pull input configured. Unified Agent periodically polls the third-party application directly, if the application supports metrics in the Prometheus format, or polls a special export application that provides integration with Prometheus.

You can find an example of how to configure HAProxy metrics delivery in Delivering metrics from HAProxy and other third-party applications.

How do I deliver metrics from hosts outside Yandex Cloud?How do I deliver metrics from hosts outside Yandex Cloud?

Configure authorized keys of service accounts to deliver metrics to Monitoring through Unified Agent.

You can also use this method to send metrics from Yandex Cloud VMs without a linked service account.

How do I view a list of metrics?How do I view a list of metrics?

To do this, open Metric Explorer and select the desired service.

To get a complete list of metrics belonging to a specific cloud and folder, follow this guide on how to make API requests.

Can I export metrics in Prometheus format?Can I export metrics in Prometheus format?

To export metrics in Prometheus format, use the prometheusMetrics method.

See an example of how to set up metrics export in the documentation.

Metrics and units of measureMetrics and units of measure

What does the dot mean?What does the dot mean?

It separates the integer part from the fractional part of a number. For example, the value 123.456 should be interpreted as 123 and 456 thousandths.

Why does cpu_usage exceed 100%?Why does cpu_usage exceed 100%?

You're using a VM with guaranteed vCPU performance under 100%. For example, 20% of vCPU performance is guaranteed. For the monitoring system, this level is taken as 100% of the expected load. If there are no "neighbors" on the physical core, your VM can be allocated up to 100% of vCPU performance, which is 5 times more than the guaranteed share. The chart can show up to 500%.

If you see the upper limit of 100% exceeded for quite a while in the charts, we recommend increasing the guaranteed vCPU performance. At any time, "neighbors" may appear on the physical core and the actual vCPU usage may drop to the guaranteed 20% (about 400 MHz for Intel Cascade Lake).

Where can I find DB metric descriptions?Where can I find DB metric descriptions?

Metrics are described in the documentation for services, such as:

• Yandex Managed Service for PostgreSQL.
• Yandex Managed Service for MySQL®.
• Yandex Managed Service for Valkey™.

Important metrics are displayed on each service's dashboard. For a full list of metrics for each service, see Metric Explorer. To export a full list of metrics, follow this guide.

How do I set up units on a chart?How do I set up units on a chart?

Units are set up in the Axes tab when editing a chart. You can select the unit type, set the number format, scale (linear or logarithmic), and the maximum and minimum values.

How are maximum/peak values displayed on a chart?How are maximum/peak values displayed on a chart?

Monitoring receives metrics for a certain period of time in groups and averages the values when creating a point. When choosing a shorter time interval, the number of values in the sample increases and the peak values on the chart may grow.

This also works in the opposite direction: when choosing a larger time interval, the point peak values are smoothed out and may not be visible on the chart.

The chart is missing or looks cropped. Why?The chart is missing or looks cropped. Why?

The data could fall outside the established chart boundaries. Check the scale settings and the maximum and minimum values in the Axes tab.

Is there a description of functions in the query language?Is there a description of functions in the query language?

You can find the function description in the Reference.

You can test the functions in the Metric Explorer section.

Alerts/notificationsAlerts/notifications

How can I reduce the number of false alerts?How can I reduce the number of false alerts?

You can increase the evaluation window or change the aggregation function in the alert parameters.

What value should I specify in a threshold?What value should I specify in a threshold?

The threshold value is set in the units of the selected metric.

For example, the value of a database cluster's free disk space threshold is specified in bytes. It is convenient to set a threshold by moving the slider on the chart with real data. Make sure the threshold value for a Warning does not exceed that for an Alarm.

How do I set an alert recipient?How do I set an alert recipient?

The following methods are supported in the notification channel:

• Email.
• SMS.
• Push notifications (in the Yandex Cloud mobile app).
• Telegram.

Specify a Yandex Cloud account as the recipient of email notifications. This can be the owner's account or an additional account created specifically for this purpose, such as a federated account or Yandex ID.

A user does not receive any alerts. Why is that?A user does not receive any alerts. Why is that?

Check the mailbox and phone number settings in the user account's notification channel.

To get notifications, the user must:

• Have the monitoring.viewer role for the folder the alert is configured in.
• In the management console settings:
  • Enable Monitoring.
  • Specify an e-mail address, phone number, and Telegram account or group.