Google Cloud Run

문서 > 통합 > Google Cloud Run

개요

Cloud Run은 HTTP 요청으로 호출 가능한 스테이트리스(Stateless) 컨테이너를 실행할 수 있게 해주는 관리형 컴퓨팅 플랫폼입니다.

이 통합을 사용하고 컨테이너를 계측하여 Datadog의 모든 Cloud Run 메트릭, 트레이스, 로그를 확인하세요.

Cloud Run for Anthos에 대한 자세한 내용은 Google Cloud Run for Anthos 문서를 참조하세요.

설정

메트릭 수집

설치

기본 메트릭 수집을 시작하려면 Google Cloud Platform 통합을 설정하세요. 커스텀 메트릭을 설정하려면 서버리스 문서를 참조하세요.

로그 수집

통합

Google Cloud Run은 감사 로그도 노출합니다. Google Cloud Run 로그는 Google Cloud Logging으로 수집되어 Cloud Pub/Sub 주제를 통해 Dataflow 작업으로 전송됩니다. 아직 설정하지 않았다면 Datadog Dataflow 템플릿을 사용하여 로깅을 설정하세요.

이 작업이 완료되면 Google Cloud Logging에서 Google Cloud Run 로그를 Pub/Sub 주제로 내보냅니다.

Google Cloud Logging 페이지로 이동하여 Google Cloud Run 로그를 필터링하세요.
Create Sink를 클릭하고 그에 따라 싱크 이름을 지정합니다.
“Cloud Pub/Sub"를 대상으로 선택하고 해당 목적으로 생성된 Pub/Sub 주제를 선택합니다. 참고: Pub/Sub 주제는 다른 프로젝트에 있을 수 있습니다.
Create를 클릭하고 확인 메시지가 나타날 때까지 기다립니다.

직접 로깅

Cloud Run 서비스에서 Datadog에 직접 애플리케이션을 로깅하는 방법은 서버리스 문서를 참조하세요.

트레이싱

완전 관리형 Google Cloud Run에 대한 전문적인 Agent 설정 지침은 서버리스 문서에서 확인하세요.

수집한 데이터

메트릭

gcp.run.container.billable_instance_time (rate)	Billable time aggregated from all container instances of the revision (ms/s). Shown as millisecond
gcp.run.container.completed_probe_attempt_count (count)	Number of completed health check probe attempts and their results.
gcp.run.container.completed_probe_count (count)	Number of completed health check probes and their results.
gcp.run.container.containers (gauge)	Number of container instances that exist, broken down by state.
gcp.run.container.cpu.allocation_time (rate)	Container CPU allocation of the revision in seconds. Shown as core
gcp.run.container.cpu.usage.avg (gauge)	The average actual container CPU usage in CPU seconds broken down by the metric field, container name. Shown as second
gcp.run.container.cpu.usage.samplecount (gauge)	The sample count for actual container CPU usage in CPU seconds broken down by the metric field, container name. Shown as second
gcp.run.container.cpu.usage.sumsqdev (gauge)	The sum of squared deviation for actual container CPU usage in CPU seconds broken down by the metric field, container name. Shown as second
gcp.run.container.cpu.utilizations.avg (gauge)	The average distribution of container CPU utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.cpu.utilizations.p95 (gauge)	The 95th percentile distribution of container CPU utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.cpu.utilizations.p99 (gauge)	The 99th percentile distribution of container CPU utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.cpu.utilizations.samplecount (count)	Sample count of the distribution of service request times in milliseconds. Shown as fraction
gcp.run.container.gpu.memory_usages.avg (gauge)	The average container GPU memory usage distribution across all container instances. Shown as byte
gcp.run.container.gpu.memory_usages.samplecount (gauge)	The sample count for container GPU memory usage distribution across all container instances. Shown as byte
gcp.run.container.gpu.memory_usages.sumsqdev (gauge)	The sum of squared deviation for container GPU memory usage distribution across all container instances. Shown as byte
gcp.run.container.gpu.memory_utilizations.avg (gauge)	The average container GPU memory utilization distribution across all container instances.
gcp.run.container.gpu.memory_utilizations.samplecount (gauge)	The sample count for container GPU memory utilization distribution across all container instances.
gcp.run.container.gpu.memory_utilizations.sumsqdev (gauge)	The sum of squared deviation for container GPU memory utilization distribution across all container instances.
gcp.run.container.gpu.utilizations.avg (gauge)	The average container GPU utilization distribution across all container instances.
gcp.run.container.gpu.utilizations.samplecount (gauge)	The sample count for container GPU utilization distribution across all container instances.
gcp.run.container.gpu.utilizations.sumsqdev (gauge)	The sum of squared deviation for container GPU utilization distribution across all container instances.
gcp.run.container.instance_count (gauge)	The number of container instances that exist, broken down by state. Shown as container
gcp.run.container.max_request_concurrencies.avg (gauge)	Average of the maximum number of concurrent requests being served by each container instance over a minute. Shown as request
gcp.run.container.max_request_concurrencies.p95 (gauge)	95th percentile distribution of the maximum number of concurrent requests being served by each container instance over a minute. Shown as request
gcp.run.container.max_request_concurrencies.p99 (gauge)	99th percentile distribution of the maximum number of concurrent requests being served by each container instance over a minute. Shown as request
gcp.run.container.max_request_concurrencies.samplecount (count)	Sample count of the distribution of the maximum number of concurrent requests being served by each container instance over a minute. Shown as request
gcp.run.container.memory.allocation_time (rate)	Container memory allocation of the revision in Gigabyte-seconds. Shown as gibibyte
gcp.run.container.memory.usage.avg (gauge)	The average actual container memory usage in bytes broken down by the metric field, container name. Shown as byte
gcp.run.container.memory.usage.samplecount (gauge)	The sample count for actual container memory usage in bytes broken down by the metric field, container name. Shown as byte
gcp.run.container.memory.usage.sumsqdev (gauge)	The sum of squared deviation for actual container memory usage in bytes broken down by the metric field, container name. Shown as byte
gcp.run.container.memory.utilizations.avg (gauge)	Average of the container memory utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.memory.utilizations.p95 (gauge)	95th percentile distribution of the container memory utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.memory.utilizations.p99 (gauge)	99th percentile distribution of the container memory utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.memory.utilizations.samplecount (count)	Sample count of the container memory utilization distribution across all container instances of the revision. Shown as fraction
gcp.run.container.network.received_bytes_count (count)	The incoming socket and HTTP response traffic of revision, in bytes. Shown as byte
gcp.run.container.network.sent_bytes_count (count)	The outgoing socket and HTTP response traffic of revision, in bytes. Shown as byte
gcp.run.container.network.throttled_inbound_bytes_count (count)	Inbound bytes dropped due to network throttling. Shown as byte
gcp.run.container.network.throttled_inbound_packets_count (count)	Inbound packets dropped due to network throttling. Shown as byte
gcp.run.container.network.throttled_outbound_bytes_count (count)	Outbound bytes dropped due to network throttling. Shown as byte
gcp.run.container.network.throttled_outbound_packets_count (count)	Outbound packets dropped due to network throttling. Shown as byte
gcp.run.container.probe_attempt_latencies.avg (count)	The average distribution of time spent running a single probe attempt before success or failure in milliseconds. Shown as millisecond
gcp.run.container.probe_attempt_latencies.samplecount (count)	The sample count for distribution of time spent running a single probe attempt before success or failure in milliseconds. Shown as millisecond
gcp.run.container.probe_attempt_latencies.sumsqdev (count)	The sum of squared deviation for distribution of time spent running a single probe attempt before success or failure in milliseconds. Shown as millisecond
gcp.run.container.probe_latencies.avg (count)	The average distribution of time spent running a probe before success or failure in milliseconds. Shown as millisecond
gcp.run.container.probe_latencies.samplecount (count)	The sample count for distribution of time spent running a probe before success or failure in milliseconds. Shown as millisecond
gcp.run.container.probe_latencies.sumsqdev (count)	The sum of squared deviation for distribution of time spent running a probe before success or failure in milliseconds. Shown as millisecond
gcp.run.container.startup_latencies.avg (count)	The average distribution of time spent starting a new container instance in milliseconds. Shown as millisecond
gcp.run.container.startup_latencies.samplecount (count)	The sample count for distribution of time spent starting a new container instance in milliseconds. Shown as millisecond
gcp.run.container.startup_latencies.sumsqdev (count)	The sum of squared deviation for distribution of time spent starting a new container instance in milliseconds. Shown as millisecond
gcp.run.infrastructure.cloudsql.connection_latencies.avg (count)	The average distribution of latency in microseconds for connections originating from Cloud Run to CloudSQL. Shown as microsecond
gcp.run.infrastructure.cloudsql.connection_latencies.samplecount (count)	The sample count for distribution of latency in microseconds for connections originating from Cloud Run to CloudSQL. Shown as microsecond
gcp.run.infrastructure.cloudsql.connection_latencies.sumsqdev (count)	The sum of squared deviation for distribution of latency in microseconds for connections originating from Cloud Run to CloudSQL. Shown as microsecond
gcp.run.infrastructure.cloudsql.connection_refused_count (count)	Total number of connections refused originating from Cloud Run to CloudSQL.
gcp.run.infrastructure.cloudsql.connection_request_count (count)	Total number of connection requests originating from Cloud Run to CloudSQL.
gcp.run.infrastructure.cloudsql.received_bytes_count (count)	Number of bytes received by Cloud Run from CloudSQL over the network. Shown as byte
gcp.run.infrastructure.cloudsql.sent_bytes_count (count)	Number of bytes sent by Cloud Run to CloudSQL over the network. Shown as byte
gcp.run.job.completed_execution_count (count)	Number of completed job executions and their result.
gcp.run.job.completed_task_attempt_count (count)	Number of completed task attempts and its corresponding exit result.
gcp.run.job.running_executions (gauge)	Number of running job executions.
gcp.run.job.running_task_attempts (gauge)	Number of running task attempts.
gcp.run.pending_queue.pending_requests (gauge)	Number of pending requests.
gcp.run.request_count (count)	The number of service requests. Shown as request
gcp.run.request_latencies.avg (gauge)	Average distribution of service request times in milliseconds. Shown as millisecond
gcp.run.request_latencies.p95 (gauge)	The 95th percentile distribution of service request times in milliseconds. Shown as millisecond
gcp.run.request_latencies.p99 (gauge)	The 99th percentile distribution of service request times in milliseconds. Shown as millisecond
gcp.run.request_latencies.samplecount (count)	Sample count of the distribution of service request times in milliseconds. Shown as millisecond
gcp.run.request_latencies.sumsqdev (gauge)	Sum of squared deviation of the distribution of service request times in milliseconds. Shown as millisecond