Monitoring

This page discusses options for monitoring requests to Cloud Storage, including tracking errors, request rates, and data usage.

Monitoring options

Within the Google Cloud console, there are several levels of the resource hierarchy at which you can monitor your Cloud Storage requests:

Individual buckets: Within each Bucket details page, there is an Observability tab that reports metrics for that particular bucket.
All buckets: Each project has a Cloud Storage Monitoring page, accessible in the left pane for Cloud Storage. This page can also be directly accessed using the following URL:
```
https://console.cloud.google.com/storage/monitoring?project=PROJECT_ID
```
Where PROJECT_ID is the ID of the project whose metrics you want to monitor.
All Google Cloud resources: For monitoring across all services and resources in your Google Cloud project, folder, or organization, use the Cloud Monitoring service in the Google Cloud console. This service is available at https://console.cloud.google.com/monitoring.

Default monitoring

By default, the following graphs appear in your individual bucket monitoring and your Cloud Storage Monitoring page:

Graph	Description
Server Error Rate	Requests with 5xx error codes. These errors are generally caused by issues internal to Google Cloud and are retryable. 5xx error codes are applicable to the Cloud Storage SLA.
Client Error Rate	Requests with 4xx error codes or a 304 status code with a `FAILED_PRECONDITION` error. These errors indicate changes that need to be made by or made for requesters, such as granting requesters sufficient permission to access the relevant Cloud Storage resources or scaling back requests to prevent exhausting quotas.
Read Errors	Object reading errors, specifically for errors with an `INTERNAL`, `UNAVAILABLE`, or `DEADLINE_EXCEEDED` response code. Errors related to requests for bucket metadata, object listings, or object metadata are not included in this graph. For single-bucket monitoring, errors are aggregated for the entire bucket. For Cloud Storage Monitoring, errors are aggregated for each bucket, and the 5 buckets with the most read errors are displayed by default. The graph can be customized to display additional buckets.
Write Errors	Object write errors, specifically for errors with an `INTERNAL`, `UNAVAILABLE`, or `DEADLINE_EXCEEDED` response code. Errors related to requests for object rewrites, object compositions, metadata updates, and XML API multipart uploads are not included in this graph. For single-bucket monitoring, errors are aggregated for the entire bucket. For Cloud Storage Monitoring, errors are aggregated for each bucket, and the 5 buckets with the most write errors are displayed by default. The graph can be customized to display additional buckets.
Total Read/List/Get Request Count	For single-bucket monitoring, the total number of get requests made to the bucket. For Cloud Storage Monitoring, the total number of get requests made to different locations. Counts are grouped by method type, such as requests for bucket metadata, requests for object metadata, requests for object data, and requests to list objects.
Total Write Request Count	For single-bucket monitoring, the total number of object write requests made to the bucket. For Cloud Storage Monitoring, the total number of object write requests made to different locations. Requests for object rewrites, object compositions, metadata updates, and XML API multipart uploads are not included in this graph.
Data Ingress Rate over the Network	For single-bucket monitoring, the rate of received bytes by the bucket. For Cloud Storage Monitoring, the rate of received bytes by all buckets in each location.
Data Egress Rate over the Network	For single-bucket monitoring, the rate of bytes sent by the bucket. For Cloud Storage Monitoring, the rate of bytes sent by all buckets in each location.

Graphs specific to the Observability page

By default, the Observability tab available for each bucket also contains the following monitoring graphs:

Graph	Description
Percent of minutes out of RPO	Applicable to buckets in dual-regions and multi-regions. The percentage of minutes during which one or more objects took longer than their RPO (Recovery Point Objective) time to be replicated between geographically distinct locations. Metrics are aggregated over the previous 30 days. There is a delay, typically several hours, between when a replication occurs and when monitoring reports the associated metrics. For more information, including RPO times for different replication options, see Performance monitoring.
Percent of objects out of target	Applicable to buckets in dual-regions and multi-regions. The percentage of objects that were not replicated between geographically distinct locations within the RPO (Recovery Point Objective) time. Metrics are aggregated over the previous 30 days. When tracking default replication metrics, this graph also includes a line that measures the percentage of objects that were not replicated between geographically distinct locations within the 1 hour design target. There is a delay, typically several hours, between when the replication occurs and when monitoring reports the associated metrics. For more information, including RPO times for different replication options, see Performance monitoring.
Bucket meeting RPO	Applicable to buckets in dual-regions and multi-regions. A boolean value indicating whether all objects in the bucket are meeting RPO or not. For more information, including RPO times for different replication options, see Performance monitoring.
Maximum delay in turbo replication	Applicable to buckets in dual-regions with turbo replication. The age, in seconds, of the oldest unreplicated object in the bucket.

Graphs specific to the Monitoring page

By default, the Cloud Storage Monitoring page also contains the following monitoring graphs:

Graph	Description
Google Egress Bandwidth Usage	The amount of data egress from Cloud Storage to other Google Cloud services. This graph only tracks egress from buckets located in regions, and it groups data by region. For more information, see Bandwidth usage in Cloud Storage.

You can create your own dashboard layouts and customize the data you collect by clicking the Customize dashboard button, which appears in both individual bucket monitoring tabs and in your Cloud Storage Monitoring page. For more formation, see Manage custom dashboards.