Track storage health alerts for projects

Platform Administrators (PA) and Application Operators (AO) access metrics in the monitoring instance to monitor usage and capacity in the system. These metrics have associated labels to differentiate measured characteristics, such as the stock-keeping unit (SKU) name, tenant type, and bucket.

According to your role, you have access to the following metrics from the monitoring dashboard:

Monitoring metrics
Name Description Labels Roles
obj_tenant_used_capacity_bytes_count Number of bytes consumed in object storage.
  • Bucket
  • SKU ID
  • SKU name
  • Tenant type
 PA and AO
obj_tenant_objects_count Number of objects in object storage.
  • Bucket
  • SKU ID
  • SKU name
  • Tenant type
 PA and AO
obj_tenant_quota_bytes Quota of a tenant in object storage.
  • SKU ID
  • SKU name
  • Tenant type
 PA
obj_proxy_servehttp_request_total A counter that tracks the total number of S3 requests to object storage.
  • operation_type
  • result
 PA and AO
obj_proxy_servehttp_first_byte_latency_milliseconds The time it takes for a single operation or request to complete.
  • operation_type
  • result
 PA and AO

The following table contains some examples of how to use the previous monitoring metrics:

Query examples
Query Description Roles
(sum by (tenant_type, sku_id) (obj_tenant_used_capacity_bytes_count) / sum by (tenant_type, sku_id) (obj_tenant_quota_bytes)) Percentage of used capacity per tenant. PA
obj_tenant_used_capacity_bytes_count Bytes used in each bucket. PA and AO
obj_tenant_objects_count Objects used in each bucket. PA and AO
obj_tenant_used_capacity_bytes_count{bucket=FULLY_QUALIFIED_BUCKET_NAME} Bytes used in a particular bucket. PA and AO
histogram_quantile(0.95, sum(rate(obj_proxy_servehttp_first_byte_latency_milliseconds{operation_type="read_object", result="success"}[5m])) by (le)) Calculate 95th percentile s3_get_object_latency_milliseconds. PA and AO
sum by (operation_type) (increase(obj_proxy_servehttp_request_total{operation_type="(^list_bucket$)", result=(^success$)|(^user_error$)}[5m])) Calculate total successful `list_buckets` successful requests over 5 minutes. PA and AO