Monitoring for Code Engine

Get insight into the performance of your workloads that are deployed with IBM Cloud® Code Engine. Metrics can help you find bottlenecks or predict possible production problems.

You can use the IBM Cloud® Monitoring service to monitor your Code Engine workloads. Code Engine forwards selected information about your workloads to Monitoring so that you can monitor specific metrics such as requests, revisions, and duration.

Set up your IBM Cloud Monitoring service instance

To set up your Code Engine customer metrics dashboards in Monitoring, you must create a Monitoring instance and then enable Platform Metrics in the same region as the Code Engine projects that you want to monitor. If you have deployments in more than one region, you must provision Monitoring and enable platform metrics for each region. For more information, see Monitoring Getting started tutorial.

To set up Monitoring,

  1. From the IBM Cloud navigation menu, select Observability.
  2. Select Monitoring.
  3. Either use an existing Monitoring service instance or create a new one.
  4. After the instance is ready, click Configure platform metrics to enable platform metrics.
  5. Select a region and then a Monitoring instance from that region. If you have deployments in more than one region, you must provision Monitoring and enable platform metrics for each region.
  6. Enable monitoring in your project settings. For more information, see Configuring project settings for observability.

You can also start monitoring from your Code Engine dashboard by selecting Launch Monitoring.

You can adapt the dashboards by creating a copy of a dashboard template and by editing it.

Configuring Project Settings for Observability

You can configure Monitoring Settings for your project to control the flow of metrics. By enabling or disabling specific metrics at the project level, you can manage which performance metrics are collected and forwarded to your monitoring service instance.

The rollout plan for these capabilities includes a pre-configuration phase. During this phase, settings are visible and editable but not yet active. For details, see Upcoming changes to IBM Cloud Code Engine monitoring capabilities.

Metric Selection

When you configure monitoring settings using the console or CLI, you can configure metric selection with the following options:

  • All: All metrics are enabled for the project. Code Engine collects and forwards application, function, job, compute, and GPU Metrics to your monitoring instance.
  • None: Metric collection is disabled for the project. Code Engine does not scrape or forward any metrics.
  • Custom: Granularly configure which metric categories are collected and forwarded. You can individually enable (On) or disable (Off) the following metrics:

You can configure monitoring settings from the console or by using the CLI.

Configuring Monitoring Settings from the Console

Use the Code Engine console to configure project-wide monitoring settings.

Before you can customize workload metrics, you must first connect an IBM Cloud Monitoring instance to enable platform metrics for the project's region.

  1. Go to the Code Engine Projects page in the console.
  2. Select your project.
  3. Select Project settings > Observability.
  4. In the Platform Monitoring section, verify that platform metrics are enabled. If not enabled, click Add monitoring to connect a monitoring instance in the project's region.
  5. In the Metrics selection section, select your preferred mode:
    • All: Collect and forward all application, function, job, compute, and GPU Metrics.
    • None: Disable metric forwarding for all workloads in this project.
    • Custom: Granularly enable or disable specific metric categories.
  6. Click Save.

Configuring Monitoring Settings with the CLI

Use CLI commands to view and update project-wide monitoring settings.

  1. To retrieve the current monitoring settings for your project, run the ibmcloud ce settings monitoring get command. For a complete list of all available command options, see the CLI reference.

    ibmcloud ce settings monitoring get
    

    By default, the command returns the following output, where standard metrics are set to Custom:

    Enable metrics:  custom
    
    Standard metrics:
      Functions:     on
      Applications:  on
      Jobs:          on
      Compute:       off
      GPU:           off
    
  2. To change the monitoring settings for your project, run the ibmcloud ce settings monitoring update command with one of the following options. For a complete list of all available command options, see the CLI reference.

    • To set metric selection to all, run the command with the --standard-metrics all option.

      ibmcloud ce settings monitoring update --standard-metrics all
      
      Updating project monitoring settings...
      OK
      
      Project Name:  myproject
      Project ID:    01234567-abcd-abcd-abcd-abcdabcd1111
      
      Enable metrics:  all
      
      All standard metrics are enabled for this project.
      
      Standard metrics:
        Functions
        Applications
        Jobs
        Compute
        GPU
      
    • To disable metric selection, run the command with the --standard-metrics none option.

      ibmcloud ce settings monitoring update --standard-metrics none
      
      Updating project monitoring settings...
      OK
      
      Project Name:  myproject
      Project ID:    01234567-abcd-abcd-abcd-abcdabcd1111
      
      Enable metrics:  none
      
      Standard metrics are disabled for this project.
      
    • To enable custom (granular) metric selection, run the command with the --standard-metrics custom option. When you set metric selection to custom, the granular settings take precedence on the basis of being set to on or off.

      ibmcloud ce settings monitoring update --standard-metrics custom
      
      Updating project monitoring settings...
      OK
      
      Project Name:  myproject
      Project ID:    01234567-abcd-abcd-abcd-abcdabcd1111
      
      Enable metrics:  custom
      
      Standard metrics:
        Functions:     on
        Applications:  on
        Jobs:          on
        Compute:       on
        GPU:           off
      

Configure Application Metrics

When the Applications Metrics setting is set to On, Code Engine collects and forwards scaling, traffic, and concurrency metrics for your application workloads.

The following table describes the standard Application Metrics:

Application Metrics
Metric Description
ibm_codeengine_application_actual_instances Number of pods that are currently allocated
ibm_codeengine_application_requested_instances Number of pods requested
ibm_codeengine_application_not_ready_instances Number of pods that are allocated but not yet ready to receive requests
ibm_codeengine_application_pending_instances Number of pods that are pending allocation
ibm_codeengine_application_desired_instances Number of pods that are desired by the autoscaler
ibm_codeengine_application_terminating_instances Number of pods that are terminating
ibm_codeengine_application_requests_total Total number of requests that are received by the application
ibm_codeengine_application_revision_count Total number of revisions for the application
ibm_codeengine_application_service_count Total number of services
ibm_codeengine_application_route_count Total number of routes
ibm_codeengine_application_target_concurrency_per_pod Configured target concurrency per pod
ibm_codeengine_application_panic_request_concurrency Target concurrency during panic mode
ibm_codeengine_application_stable_request_concurrency Target concurrency during stable mode
ibm_codeengine_application_panic_mode Specifies whether the autoscaler is operating in panic mode

Configure Function Metrics

When the Functions Metrics setting is set to On, Code Engine collects and forwards invocation, execution duration, and status metrics for your functions.

The following table describes the standard Function Metrics:

Function Metrics
Metric Description
ibm_codeengine_functions_activations Total number of function invocations
ibm_codeengine_functions_exceeded_limit Total number of function requests that exceed execution limits
ibm_codeengine_functions_execution_time_count Count of recorded execution time measurements
ibm_codeengine_functions_execution_time_sum Cumulative execution time (in milliseconds)
ibm_codeengine_functions_init_time_count Count of initialization time measurements
ibm_codeengine_functions_init_time_sum Cumulative initialization time (in milliseconds)
ibm_codeengine_functions_status_success Number of successfully executed function invocations
ibm_codeengine_functions_status_error_application Number of function invocations that fail with an application error
ibm_codeengine_functions_status_error_developer Number of invoked functions that fail with an error in your code

Configure Job Metrics

When the Jobs Metrics setting is set to On, Code Engine collects and forwards workload execution metrics for your batch jobs.

The following table describes the standard Job Metrics:

Job Metrics
Metric Description
ibm_codeengine_jobruns Total number of job runs in your project

Configure Compute Metrics

When the Compute Metrics setting is set to On, Code Engine collects and forwards container-level CPU and memory usage metrics for the project's instances.

The following table describes the Compute Metrics for applications, jobs, and builds:

Compute Metrics
Metric Description
ibm_codeengine_instance_cpu_usage_millicores Actual CPU usage of the instance in millicores
ibm_codeengine_instance_cpu_limit_millicores CPU limit configured for the instance in millicores
ibm_codeengine_instance_memory_usage_bytes Actual memory usage of the instance in bytes
ibm_codeengine_instance_memory_limit_bytes Memory limit configured for the instance in bytes

For fleets, enabling Compute Metrics collects container-level metrics from the Sysdig metrics library. For more information, see the Sysdig container metrics library.

Configure GPU Metrics for fleets

When the GPU Metrics setting is set to On, Code Engine collects and forwards NVIDIA GPU utilization and hardware performance metrics for your GPU-enabled fleet workers.

To enable GPU Metrics, the Compute Metrics setting must also be set to On. The combination of Compute: Off and GPU: On is invalid and cannot be configured.

The following table describes the standard GPU Metrics:

GPU Metrics
Metric Description
ibm_codeengine_xpu_utilization_ratio GPU utilization (%)
ibm_codeengine_xpu_memory_used_bytes Framebuffer memory used (MiB)
ibm_codeengine_xpu_memory_free_bytes Framebuffer memory free (MiB)
ibm_codeengine_xpu_temperature_celsius GPU temperature (°C)
ibm_codeengine_xpu_power_usage_watts Power draw (W)
ibm_codeengine_xpu_clock_mhz Streaming multiprocessor clock (MHz)
ibm_codeengine_xpu_pcie_rx_bytes_total PCIe RX throughput (bytes/s)
ibm_codeengine_xpu_pcie_tx_bytes_total PCIe TX throughput (bytes/s)
ibm_codeengine_xpu_encoder_utilization_ratio GPU encoder utilization (%)
ibm_codeengine_xpu_decoder_utilization_ratio GPU decoder utilization (%)

Accessing your IBM Cloud Monitoring metrics

To see your Code Engine customer metrics dashboards in Monitoring:

  1. From the IBM Cloud navigation menu, select Observability.
  2. Select Monitoring. Then select Instances.
  3. Click Dashboard in the View column of your monitoring instance to open the dashboard for your monitoring instance.
  4. From the navigation menu, select Dashboards->IBM->IBM Code Engine Project Overview, Dashboards->IBM->IBM Code Engine Functions Overview, Dashboards->IBM->IBM Code Engine Instance Resource Overview, or Dashboards->IBM->IBM Code Engine Fleet GPU Monitoring. If you don't see the Code Engine dashboard in the menu, you can start monitoring from your Code Engine application or job by selecting Add Monitoring.
  5. Select the 10M timeline or greater. Because Platform Metrics data has a 1 minute granularity, the first timeline that shows metrics is the 10M timeline.

You can also start the Monitoring dashboard at any time by selecting Monitoring from the Code Engine Action menu.

When you use the IBM Cloud Monitoring service to monitor Code Engine workloads, delays can occur before the data is processed and available in Monitoring. For example, right after you create your monitoring instance, it takes a few minutes before meaningful monitoring data is displayed. Be aware that it might take around 5 minutes for your monitoring data to show in Monitoring.

ibm_codeengine_application_actual_instances

Configure Application Metrics

Number of pods that are currently allocated.

Use this metric to observe the behavior of automatic application scaling and to observe how many application instances are running. Monitor this value to determine if you need to adjust the configurable values for your application.

The number of running instances of an app are automatically scaled up or down based on configuration settings and your workloads. Application scaling is configurable. See Configuring application scaling and be aware of Application defaults and limits.

Number of pods that are currently allocated
Metadata Description
Metric Name ibm_codeengine_application_actual_instances
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_requested_instances

Configure Application Metrics

Number of pods that the autoscaler requested from Kubernetes.

Use this metric to observe the behavior of automatic application scaling. Monitor this value to determine if you need to adjust the configurable values for your application.

The number of running instances of an app are automatically scaled up or down based on configuration settings and your workloads. Application scaling is configurable. See Configuring application scaling and be aware of Application defaults and limits.

Number of pods that the autoscaler requested
Metadata Description
Metric Name ibm_codeengine_application_requested_instances
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_not_ready_instances

Configure Application Metrics

Number of pods that are not ready currently.

Use this metric to observe if there are issues with your running application.

Application instances that are in not ready state cannot serve requests. This not ready state indicates that something is preventing the app from becoming ready.

Number of pods that are not ready currently
Metadata Description
Metric Name ibm_codeengine_application_not_ready_instances
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_pending_instances

Configure Application Metrics

Number of pods that are pending currently.

Use this metric to observe if there are issues with your running application.

You might observe pending applications when you observe automatic application scaling. Application instances that are pending are waiting to be scheduled. If your application remains in a pending state, perhaps the application cannot start because of insufficient resources, such as memory or CPU. See Supported memory and CPU combinations.

Number of pods that are pending currently
Metadata Description
Metric Name ibm_codeengine_application_pending_instances
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_desired_instances

Configure Application Metrics

Number of pods that the autoscaler wants to allocate.

Use this metric to observe the behavior of automatic application scaling. Monitor this value to determine if you need to adjust the configurable values for your application.

The number of running instances of an app are automatically scaled up or down based on configuration settings and your workloads. Application scaling is configurable. See Configuring application scaling and be aware of Application defaults and limits.

Number of pods that the autoscaler wants to allocate
Metadata Description
Metric Name ibm_codeengine_application_desired_instances
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_terminating_instances

Configure Application Metrics

Number of pods that are terminating currently.

Use this metric to observe the behavior of automatic application scaling. Monitor this value to observe if an application terminates when applications are scaled down.

The number of running instances of an app are automatically scaled up or down based on configuration settings and your workloads. Application scaling is configurable. See Configuring application scaling and be aware of Application defaults and limits.

Number of pods that are terminating currently
Metadata Description
Metric Name ibm_codeengine_application_terminating_instances
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_requests_total

Configure Application Metrics

Total number of HTTPS requests to the application.

Use this metric to monitor the number of HTTPS requests that are received by your application.

Total number of HTTPS requests to the application
Metadata Description
Metric Name ibm_codeengine_application_requests_total
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name, http status code

ibm_codeengine_application_revision_count

Configure Application Metrics

Number of revisions per application.

Use this metric to observe the number of revisions per application.

An application contains one or more revisions. Each update of an application configuration property creates a new revision of the application. Code Engine has a quota for the number of apps and app revisions in a project. For more information about limits for projects, see Project quotas.

Number of revisions per application
Metadata Description
Metric Name ibm_codeengine_application_revision_count
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name

ibm_codeengine_application_service_count

Configure Application Metrics

Number of applications per project.

Use this metric to observe how many applications are in your project.

Code Engine has a quota for the number of apps and app revisions in a project. For more information about limits for projects, see Project quotas.

Number of applications per project
Metadata Description
Metric Name ibm_codeengine_application_service_count
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name

ibm_codeengine_application_route_count

Configure Application Metrics

Number of routes per application.

Code Engine supports only 1 route per application.

Number of routes per application
Metadata Description
Metric Name ibm_codeengine_application_route_count
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name

ibm_codeengine_application_target_concurrency_per_pod

Configure Application Metrics

The number of concurrent requests that you want for each pod.

Use this metric to observe the target concurrency value that is set for your application.

Target concurrency for application scaling is configurable. See Configuring application scaling.

The number of concurrent requests that you want for each pod
Metadata Description
Metric Name ibm_codeengine_application_target_concurrency_per_pod
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_panic_request_concurrency

Configure Application Metrics

Average of requests count over the panic window.

Use this metric to observe the behavior of automatic application scaling. Monitor this value to observe if the application traffic arrives in bursts or if the traffic is steady.

While stable mode is used for general operations, panic mode has a much shorter window, and is used to quickly scale up an application revision if a burst of traffic occurs.

Average of requests count over the panic window
Metadata Description
Metric Name ibm_codeengine_application_panic_request_concurrency
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_stable_request_concurrency

Configure Application Metrics

Average of requests count over the stable window.

Use this metric to observe the behavior of automatic application scaling. Monitor this value to observe if the application traffic arrives in bursts or if the traffic is steady.

While stable mode is used for general operations, panic mode has a much shorter window, and is used to quickly scale up an application revision if a burst of traffic occurs.

Average of requests count over the stable window
Metadata Description
Metric Name ibm_codeengine_application_stable_request_concurrency
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_application_panic_mode

Configure Application Metrics

Specifies whether panic mode is enabled.

Use this metric to observe the behavior of automatic application scaling. Monitor this value to observe whether panic mode is used.

While stable mode is used for general operations, panic mode has a much shorter window, and is used to quickly scale up an application revision if a burst of traffic occurs.

Specifies if panic mode is enabled
Metadata Description
Metric Name ibm_codeengine_application_panic_mode
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, application name, application revision name

ibm_codeengine_functions_activations

Configure Function Metrics

Number of invoked functions.

Number of invoked functions
Metadata Description
Metric Name ibm_codeengine_functions_activations
Metric Type counter
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name, function type

ibm_codeengine_functions_exceeded_limit

Configure Function Metrics

Number of functions that are invoked after the limit for your project is exceeded. For more information about limits for projects, see Project quotas.

Number of invoked functions after exceeding project limit
Metadata Description
Metric Name ibm_codeengine_functions_exceeded_limit
Metric Type counter
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name, function type

ibm_codeengine_functions_execution_time_count

Configure Function Metrics

Number of invoked functions that run to completion.

The ibm_codeengine_functions_execution_time_count and ibm_codeengine_functions_execution_time_sum metrics are used to determine the average execution time per invoked function.

Number of invoked functions that complete execution
Metadata Description
Metric Name ibm_codeengine_functions_execution_time_count
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name

ibm_codeengine_functions_execution_time_sum

Configure Function Metrics

Sum of all execution time for invoked functions that run to completion.

The ibm_codeengine_functions_execution_time_count and ibm_codeengine_functions_execution_time_sum metrics are used to determine the average execution time per invoked function.

Sum of all execution time for invoked functions
Metadata Description
Metric Name ibm_codeengine_functions_execution_time_sum
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name

ibm_codeengine_functions_init_time_count

Configure Function Metrics

Number of functions that complete initialization such that the code is loaded in Code Engine and ready for execution.

The ibm_codeengine_functions_init_time_count and ibm_codeengine_functions_init_time_sum metrics are used to determine the average initialization time for invoked functions.

Number of functions that complete initialization
Metadata Description
Metric Name ibm_codeengine_functions_init_time_count
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name

ibm_codeengine_functions_init_time_sum

Configure Function Metrics

Sum of all initialization time for invoked functions.

The ibm_codeengine_functions_init_time_count and ibm_codeengine_functions_init_time_sum metrics are used to determine the average initialization time for invoked functions.

Sum of all initialization time for invoked functions
Metadata Description
Metric Name ibm_codeengine_functions_init_time_sum
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name

ibm_codeengine_functions_status_success

Configure Function Metrics

Number of functions that are successfully invoked. A function is invoked successfully in Code Engine after the function completes initialization, runs to completion, and your function code responds with a successful HTTP response status code.

Number of functions successfully invoked
Metadata Description
Metric Name ibm_codeengine_functions_status_success
Metric Type counter
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name, function type

ibm_codeengine_functions_status_error_application

Configure Function Metrics

Number of invoked functions that fail during Code Engine processing.

Number of invoked functions that fail within Code Engine
Metadata Description
Metric Name ibm_codeengine_functions_status_error_application
Metric Type counter
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name, function type

ibm_codeengine_functions_status_error_developer

Configure Function Metrics

Number of invoked functions that fail with an error in your code.

Number of invoked functions that fail with an error in your code
Metadata Description
Metric Name ibm_codeengine_functions_status_error_developer
Metric Type counter
Value Type none
Segment By Service instance, name of the namespace, project name, function name, function pod name, function type

ibm_codeengine_instance_cpu_usage_millicores

Configure Compute Metrics

Actual CPU usage of the instance in millicores.

Actual CPU usage of the instance in millicores
Metadata Description
Metric Name ibm_codeengine_instance_cpu_usage_millicores
Metric Type gauge
Value Type none
Segment By Project Name, Component Type, Component Name, instance

ibm_codeengine_instance_cpu_limit_millicores

Configure Compute Metrics

CPU limit configured for the instance in millicores.

CPU limit configured for the instance in millicores
Metadata Description
Metric Name ibm_codeengine_instance_cpu_limit_millicores
Metric Type gauge
Value Type none
Segment By Project Name, Component Type, Component Name, instance

ibm_codeengine_instance_memory_usage_bytes

Configure Compute Metrics

Actual memory usage of the instance in bytes.

Actual memory usage of the instance in bytes
Metadata Description
Metric Name ibm_codeengine_instance_memory_usage_bytes
Metric Type gauge
Value Type none
Segment By Project Name, Component Type, Component Name, instance

ibm_codeengine_instance_memory_limit_bytes

Configure Compute Metrics

Memory limit configured for the instance in bytes.

Memory limit configured for the instance in bytes
Metadata Description
Metric Name ibm_codeengine_instance_memory_limit_bytes
Metric Type gauge
Value Type none
Segment By Project Name, Component Type, Component Name, instance

ibm_codeengine_xpu_utilization_ratio

Configure GPU Metrics for fleets

GPU utilization ratio (%).

GPU utilization ratio
Metadata Description
Metric Name ibm_codeengine_xpu_utilization_ratio
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_memory_used_bytes

Configure GPU Metrics for fleets

Framebuffer memory used (MiB).

Framebuffer memory used
Metadata Description
Metric Name ibm_codeengine_xpu_memory_used_bytes
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_memory_free_bytes

Configure GPU Metrics for fleets

Framebuffer memory free (MiB).

Framebuffer memory free
Metadata Description
Metric Name ibm_codeengine_xpu_memory_free_bytes
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_temperature_celsius

Configure GPU Metrics for fleets

GPU temperature (°C).

GPU temperature
Metadata Description
Metric Name ibm_codeengine_xpu_temperature_celsius
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_power_usage_watts

Configure GPU Metrics for fleets

Power draw (W).

Power draw
Metadata Description
Metric Name ibm_codeengine_xpu_power_usage_watts
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_clock_mhz

Configure GPU Metrics for fleets

Streaming multiprocessor clock (MHz).

Streaming multiprocessor clock
Metadata Description
Metric Name ibm_codeengine_xpu_clock_mhz
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_pcie_rx_bytes_total

Configure GPU Metrics for fleets

PCIe RX throughput (bytes/s).

PCIe RX throughput
Metadata Description
Metric Name ibm_codeengine_xpu_pcie_rx_bytes_total
Metric Type counter
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_pcie_tx_bytes_total

Configure GPU Metrics for fleets

PCIe TX throughput (bytes/s).

PCIe TX throughput
Metadata Description
Metric Name ibm_codeengine_xpu_pcie_tx_bytes_total
Metric Type counter
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_encoder_utilization_ratio

Configure GPU Metrics for fleets

GPU encoder utilization (%).

GPU encoder utilization
Metadata Description
Metric Name ibm_codeengine_xpu_encoder_utilization_ratio
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_xpu_decoder_utilization_ratio

Configure GPU Metrics for fleets

GPU decoder utilization (%).

GPU decoder utilization
Metadata Description
Metric Name ibm_codeengine_xpu_decoder_utilization_ratio
Metric Type gauge
Value Type none
Segment By project name, region, fleet name, worker name, device, model name, driver version

ibm_codeengine_jobruns

Configure Job Metrics

Total number of job runs.

Use this metric to monitor how many job runs are in your project. Code Engine has a quota for the number of jobs and job runs in a project. For more information about limits for projects, see Project quotas.

Total number of job runs
Metadata Description
Metric Name ibm_codeengine_jobruns
Metric Type gauge
Value Type none
Segment By Service instance, name of the namespace, project name, job Name, jobrun condition

Attributes for segmentation

Global Attributes

The following attributes are available for segmenting all the metrics previously listed. Not applicable for fleets GPU Metrics.

Global attributes
Attribute Attribute Name Attribute Description
Cloud Type ibm_ctype The cloud type is a value of public, dedicated, or local.
Location ibm_location The location of the monitored resource, which can be a region, data center, or global.
Resource ibm_resource The resource that is measured by the service, which is typically an identifying name or GUID.
Resource Type ibm_resource_type The type of the resource that is measured by the service.
Resource group ibm_resource_group_name The resource group where the service instance was created.
Scope ibm_scope The scope is the account, organization, or space GUID associated with this metric.
Service name ibm_service_name Name of the service that is generating this metric.

More attributes

The following attributes are available for segmenting one or more attributes as described in the previous tables. See the individual metrics for segmentation options.

Segmentation options
Attribute Attribute Name Attribute Description
Name of the namespace ibm_codeengine_namespace Name of the namespace. This unique identifier is contained in the URL of your application. The default URL for applications is of the format https://<appname>.<uuid>.<region>.codeengine.appdomain.cloud where appname is the name of your application, uuid is the automatically generated unique identifier, and region is the region in which your Code Engine project resides. The UUID portion of the URL is the name of the namespace.
Service instance ibm_service_instance The service instance segment identifies the instance that the metric is associated with.
HTTP status code ibm_codeengine_status HTTP status code.
Application name ibm_codeengine_application_name Name of the application.
Application revision name ibm_codeengine_revision_name Name of the application revision.
Gateway instance ibm_codeengine_gateway_instance The gateway instance.
Job Name ibm_codeengine_job_name Name of the job that is used to create the job run.
Jobrun condition ibm_codeengine_jobrun_condition The condition of the job run.
Project name ibm_codeengine_project_name Name of the project.
Function name ibm_codeengine_function_name Name of the invoked function.
Function pod name ibm_codeengine_function_podname The name of the system pod that invoked the function.
Function type ibm_codeengine_function_type The type for how the function was invoked (sync or async).
Component type ibm_codeengine_component_type Type of the Code Engine component that produced the metric. The value is app, job, or build.
Component name ibm_codeengine_component_name Name of the Code Engine component (application name or job name) that produced the metric.
Subcomponent type ibm_codeengine_subcomponent_type Type of the Code Engine subcomponent. The value is app_revision for application pods, job_run for job pods, or build_run for build pods.
Subcomponent name ibm_codeengine_subcomponent_name Name of the Code Engine subcomponent (the revision name for application pods) that produced the metric.
Instance name ibm_codeengine_instance_name Name of the pod (instance) that produced the compute metric.
Fleet project name agent_tag_project_name Name of the project that contains the fleet.
Fleet project region agent_tag_project_region Region in which the project that contains the fleet resides.
Fleet name agent_tag_fleet Name of the fleet.
Worker name host_hostname Host name of the fleet worker.
GPU model name modelName Model name of the GPU.
GPU device device GPU device on the fleet worker.