All Products
Search
Document Center

Function Compute:Monitoring metrics

Last Updated:Apr 16, 2026

In the Function Compute console, you can view overview metrics for your Function Compute resources, and detailed metrics at the region, function, and instance levels. Each metric is identified by a MetricName parameter. This topic describes the MetricName values for different Function Compute metrics.

Resource overview metrics

Log on to the Function Compute console. On the Overview page, find the Resource Usage Statistics section to view resource overview metrics.

Resource overview metrics measure the overall resource usage and network traffic of Function Compute in a specific region or across all regions. The following table describes these metrics, which are aggregated daily or monthly.

Category

Metric name

Unit

Description

Overview

Invocations

Count

The total number of function invocations.

vCPU Usage

vCPU-second

The vCPU resources consumed by functions during invocation, calculated by multiplying a function's vCPU specification by its execution duration.

MemoryUsage

GB-second

The memory resources consumed by functions during invocation, calculated by multiplying a function's memory specification by its execution duration.

DiskUsage

GB-second

The disk resources consumed by functions during invocation, calculated by multiplying a function's disk specification by its execution duration.

InternetOut

GB

The total outbound Internet traffic generated by your functions.

GPU Usage

GB-second

The GPU resources consumed by functions during invocation, calculated by multiplying a function's GPU specification by its execution duration.

vCPU usage

Active vCPU Usage

vCPU-second

The vCPU resources consumed by active function instances, calculated by multiplying a function's vCPU specification by its execution duration.

Idle vCPU Usage

vCPU-second

The vCPU resources consumed by idle function instances, calculated by multiplying a function's vCPU specification by the instance's idle duration.

GPU usage

Active GPU Usage

GB-second

The GPU resources consumed by active function instances, calculated by multiplying a function's GPU specification by its execution duration.

Idle GPU Usage

GB-second

The GPU resources consumed by idle function instances, calculated by multiplying a function's GPU specification by the instance's idle duration.

Region-level metrics

Log on to the Function Compute console. In the left-side navigation pane, choose Advanced Features > Metrics to view region-level metrics.

Region-level metrics measure the overall usage of Function Compute resources in a specific region. The following table describes these metrics.

Category

Metric name

Unit

Description

Function execution

Invocations (RegionTotalInvocations)

Count

The total number of function invocations in a region. Data is aggregated in 1-minute or 1-hour intervals.

Errors

Server Errors (RegionServerErrors)

Count

The total number of failed invocations in a region due to a Function Compute system error. Data is aggregated in 1-minute or 1-hour intervals.

Note

This does not include invocations of HTTP-triggered functions that successfully return a 5xx status code.

Client Errors (RegionClientErrors)

Count

This is the total number of invocations in a specific region where a function is not executed or fails to execute due to a Function Compute client error and a 4xx status code is returned. The invocations are summed up at a 1-minute or 1-hour granularity. The following are some examples of these status codes.

  • 400: Invalid parameters.

  • 403: Permission denied.

  • 404: Resource not found.

  • 412: Custom runtime instance failed to start.

  • 499: Client closed the connection.

For more information, see Public error codes.

Note

For requests that result in a 412 or 499 client error, if the function has started execution and generated logs and incurred charges, you can find the invocation logs for these client error requests in the Invocation List. For more information, see View invocation logs.

Function Errors (RegionFunctionErrors)

Count

The total number of invocations in a region that failed due to an error within the function code. Data is aggregated in 1-minute or 1-hour intervals.

Throttling errors

Concurrency Throttles (RegionThrottles)

Count

The total number of invocations throttled for exceeding the concurrency limit in a region. These requests return a 429 status code. Data is aggregated in 1-minute or 1-hour intervals.

Total Instance Throttles (RegionResourceThrottles)

Count

The total number of invocations throttled for exceeding the total instance limit in a region. These requests return a 503 status code. Data is aggregated in 1-minute or 1-hour intervals.

On-demand instances

On-demand Concurrency Limit (RegionConcurrencyLimit)

Count

The on-demand instance quota for your account in a region.

On-demand Concurrent Instances (RegionConcurrentCount)

Count

The number of concurrent on-demand instances in use in a region. Data is aggregated in 1-minute or 1-hour intervals.

Provisioned instances

Provisioned Concurrent Instances (RegionProvisionedCurrentInstance)

Count

The total number of provisioned instances for all functions in a region within your account.

Function-level metrics

Log on to the Function Compute console. In the left-side navigation pane, choose Advanced Features > Metrics. In the Function Name list, click a function name to view its Function Metrics.

Function-level metrics measure the resource usage of a specific function. You can view metrics for the function itself, or for a specific function version or alias. The following table describes these metrics.

Note
  • For metrics scoped to a function version or alias, the metric name is prefixed with FunctionQualifier. For example, the metric for the total number of invocations is FunctionQualifierTotalInvocations.

  • To monitor CPU usage, memory usage, and network traffic at the function level, you must enable instance-level metrics. For more information, see Instance-level metrics.

Category

Metric name

Unit

Description

Invocations

Total Invocations (FunctionTotalInvocations)

Count

The total number of function invocations from both on-demand and provisioned instances. Data is aggregated in 1-minute or 1-hour intervals.

Provisioned Invocations (FunctionProvisionInvocations)

Count

The total number of function invocations on provisioned instances. Data is aggregated in 1-minute or 1-hour intervals.

HTTP Status Codes

HTTP 2xx Responses (FunctionHTTPStatus2xx)

Count

The number of requests that returned an HTTP 2xx status code. Data is aggregated in 1-minute, 5-minute, or 1-hour intervals.

HTTP 3xx Responses (FunctionHTTPStatus3xx)

Count

The number of requests that returned an HTTP 3xx status code. Data is aggregated in 1-minute, 5-minute, or 1-hour intervals.

HTTP 4xx Responses (FunctionHTTPStatus4xx)

Count

The number of requests that returned an HTTP 4xx status code. Data is aggregated in 1-minute, 5-minute, or 1-hour intervals.

HTTP 5xx Responses (FunctionHTTPStatus5xx)

Count

The number of requests that returned an HTTP 5xx status code. Data is aggregated in 1-minute, 5-minute, or 1-hour intervals.

Errors

Server Errors (FunctionServerErrors)

Count

The total number of failed invocations for a function due to a Function Compute system error. Data is aggregated in 1-minute or 1-hour intervals.

Note

This does not include invocations of HTTP-triggered functions that successfully return a 5xx status code.

Client Errors (FunctionClientErrors)

Count

The total number of invocations in a region where a function was not executed or did not execute successfully due to a Function Compute client error, and a 4xx status code was returned. The values are summed at a 1-minute or 1-hour granularity. Some example status codes are as follows.

  • 400: Invalid parameters.

  • 403: Permission denied.

  • 404: Resource not found.

  • 412: Custom runtime instance failed to start.

  • 499: Client closed the connection.

For more information, see Public error codes.

Note

For requests that result in a 412 or 499 client error, if the function has started execution and generated logs and incurred charges, you can find the invocation logs for these client error requests in the Invocation List. For more information, see View invocation logs.

Function Errors (FunctionFunctionErrors)

Count

The total number of invocations for a specific function that failed due to an error within the function code. Data is aggregated in 1-minute or 1-hour intervals.

Throttling errors

Concurrency Throttles (FunctionConcurrencyThrottles)

Count

The total number of invocations throttled for exceeding the function's concurrency limit. These requests return a 429 status code. Data is aggregated in 1-minute or 1-hour intervals.

Resource Throttles (FunctionResourceThrottles)

Count

The total number of throttled invocations due to the function's total instance limit being reached. These requests return a 503 status code. Data is aggregated in 1-minute or 1-hour intervals.

Function execution duration

Average Duration (FunctionAvgDuration)

ms

The average execution time of your function's code, from start to finish, excluding platform overhead. Data is aggregated as an average in 1-minute or 1-hour intervals.

P90 Duration (FunctionP90Duration)

ms

The 90th percentile of function execution time, from start to finish. This does not include platform overhead. 90% of requests complete within this duration.

P99 Duration (FunctionP99Duration)

ms

The 99th percentile of function execution time, from start to finish. This does not include platform overhead. 99% of requests complete within this duration.

Maximum Duration (FunctionMaxDuration)

ms

The maximum execution time of your function code, from start to finish. This does not include platform overhead. Data is aggregated as a maximum in 1-minute or 1-hour intervals.

End-to-end latency

Average Latency (FunctionLatencyAvg)

ms

During an invocation, this is the duration from when a function execution request arrives at the Function Compute system until it leaves the Function Compute system. This duration includes the time consumed by the platform. The average time is calculated at a 1-minute or 1-hour granularity.

Memory usage

Memory Quota (FunctionMemoryLimitMB)

MB

The configured memory limit for the function. Exceeding this limit causes an out-of-memory (OOM) error. The metric reports the maximum value across all instances in 1-minute or 1-hour intervals.

Maximum Memory Used (FunctionMaxMemoryUsage)

MB

The maximum amount of memory a function consumed during invocation. The metric reports the maximum value across all instances in 1-minute or 1-hour intervals.

Function on-demand instances

On-demand Instance Quota (FunctionOndemandInstanceQuota)

Count

The configured concurrency limit for on-demand instances of the function. This metric appears only if a limit is configured.

Active On-demand Instances (FunctionOndemandActiveInstance)

Count

The number of on-demand instances the function is using.

Function provisioned instances

Active Provisioned Instances (FunctionProvisionedCurrentInstance)

Count

The number of provisioned instances the function is using.

Asynchronous invocation processing

Enqueued Requests (FunctionEnqueueCount)

Count

The number of requests queued for asynchronous invocation. Data is aggregated in 1-minute or 1-hour intervals.

Processed Requests (FunctionDequeueCount)

Count

The number of completed asynchronous invocation requests. Data is aggregated in 1-minute or 1-hour intervals.

Note

If the number of processed requests is significantly lower than the number of enqueued requests, a request backlog can occur. To resolve this, configure an auto scaling policy based on the minimum number of instances or contact us.

Asynchronous message latency

Average Latency (FunctionAsyncMessageLatencyAvg)

ms

The average time between when an asynchronous message is queued and when its processing begins. Data is aggregated as an average in 1-minute or 1-hour intervals.

Maximum Latency (FunctionAsyncMessageLatencyMax)

ms

The maximum time between when an asynchronous message is enqueued and when it starts to be processed. Data is aggregated as a maximum in 1-minute or 1-hour intervals.

Asynchronous invocation events

Expired Dropped Events (FunctionAsyncEventExpiredDropped)

Count

The total number of timed-out requests that are dropped when a destination is configured for asynchronous invocations. Data is aggregated in 1-minute or 1-hour intervals.

Destination Errors (FunctionDestinationErrors)

Count

The number of requests that failed to reach their configured destination after function execution. Data is aggregated in 1-minute or 1-hour intervals.

Destination Succeeded (FunctionDestinationSucceeded)

Count

The number of requests successfully sent to their configured destination after function execution. Data is aggregated in 1-minute or 1-hour intervals.

Asynchronous request backlog

Backlogged Requests (FunctionAsyncMessagesBacklogV2)

Count

The total number of asynchronous requests waiting to be processed. Data is aggregated in 1-minute or 1-hour intervals.

In-process Asynchronous Requests (FunctionAsyncMessagesInProcess)

Count

The approximate number of asynchronous requests currently being processed.

Concurrent requests per instance (instance-level metrics)

Maximum Concurrent Requests (FunctionMaxConcurrentRequests)

Count

The maximum number of concurrent requests processed by a single function instance. Data is aggregated as a maximum in 1-minute or 1-hour intervals.

Average Concurrent Requests (FunctionAvgConcurrentRequests)

Count

The average number of concurrent requests processed by a single function instance. Data is aggregated as an average in 1-minute or 1-hour intervals.

vCPU usage (instance-level metrics)

vCPU Quota (FunctionvCPUQuotaCores)

vCPU

The vCPU quota for the function.

Maximum vCPU Usage (FunctionMaxvCPUCores)

vCPU

The actual number of vCPUs used by the function, where 1 represents one vCPU. The metric reports the maximum value across multiple instances in 1-minute or 1-hour intervals.

Average vCPU Usage (FunctionAvgvCPUCores)

vCPU

The actual number of vCPUs used by the function, where 1 represents one vCPU. The metric reports the average value across multiple instances in 1-minute or 1-hour intervals.

vCPU utilization (instance-level metrics)

Maximum vCPU Utilization (FunctionMaxvCPUUtilization)

%

The ratio of actual vCPUs used to the vCPU quota. The metric reports the maximum value across multiple instances in 1-minute or 1-hour intervals.

Average vCPU Utilization (FunctionAvgvCPUUtilization)

%

The ratio of actual vCPUs used to the vCPU quota. The metric reports the average value across multiple instances in 1-minute or 1-hour intervals.

Network traffic (instance-level metrics)

Inbound Traffic Rate (FunctionRXBytesPerSec)

Mbps

The rate of inbound network traffic to the function.

Outbound Traffic Rate (FunctionTXBytesPerSec)

Mbps

The rate of outbound network traffic from the function.

Memory usage (instance-level metrics)

Memory Quota (FunctionMemoryLimitMB)

MB

The configured memory limit for the function.

Note

Exceeding this quota causes an OOM error.

Maximum Memory Usage (FunctionMaxMemoryUsageMB)

MB

The actual memory used by a function instance. The metric reports the maximum value across all instances in 1-minute or 1-hour intervals.

Average Memory Usage (FunctionAvgMemoryUsageMB)

MB

The actual memory used by a function instance. The metric reports the average value across all instances in 1-minute or 1-hour intervals.

Memory utilization (instance-level metrics)

Maximum Memory Utilization (FunctionMaxMemoryUtilization)

%

The ratio of actual memory used to the memory quota. The metric reports the maximum value across multiple instances in 1-minute or 1-hour intervals.

Average Utilization

(FunctionAvgMemoryUtilization)

%

The ratio of actual memory used to the memory quota. The metric reports the average value across multiple instances in 1-minute or 1-hour intervals.

GPU memory usage (instance-level metrics)

GPU Memory Quota (FunctionGPUMemoryLimitMB)

MB

The GPU memory quota.

GPU Memory Used (FunctionGPUMaxMemoryUsage)

MB

The maximum amount of GPU memory used.

GPU memory utilization (instance-level metrics)

GPU Memory Utilization (FunctionGPUMemoryUsagePercent)

%

The GPU memory utilization.

GPU SM utilization (instance-level metrics)

SM Utilization (FunctionGPUSMPercent)

%

The GPU Streaming Multiprocessor (SM) utilization.

GPU hardware encoder utilization (instance-level metrics)

Hardware Encoder Utilization (FunctionGPUEncoderPercent)

%

The GPU hardware encoder utilization.

GPU hardware decoder utilization (instance-level metrics)

Hardware Decoder Utilization (FunctionGPUDecoderPercent)

%

The GPU hardware decoder utilization.

More information

To learn how to call the CloudMonitor API for monitoring data, see Monitoring data.