Example metrics

Suggest edits

Notice: EDB will no longer offer the Hosted Cloud Service option for new customers. However, support will remain available for existing customers of the EDB Hosted plan. The associated documentation topics will soon be removed, and updates regarding the removal process will be shared as they become available. The 'Your Cloud Account' option will continue to be available.

Here are some helpful queries for monitoring the health of your Cloud Service database. This isn't a complete list. It's intended only as a starting point.

The examples are in PromQL with variables suitable for use in a Grafana dashboard.

Status

This query shows if the pods are up as provided by Kube State metrics:

kube_pod_container_status_ready{container="postgres"}

status_example

CPU

This query shows CPU usage as provided by Kubernetes metrics:

rate(container_cpu_usage_seconds_total{container="postgres"}[5m])

cpu_usage_example

Memory

This query shows memory usage in megabytes as provided by Kubernetes metrics:

container_memory_working_set_bytes{container="postgres"}/1000000

memory_usage_example

Storage

This query shows the percentage of available storage used as provided by Kubernetes metrics:

(sum(kubelet_volume_stats_used_bytes{namespace=~"default"}) by (persistentvolumeclaim) / sum(kubelet_volume_stats_capacity_bytes{namespace=~"default"}) by (persistentvolumeclaim)) * 100

pvc_usage_example

Replication lag

This query shows the replication lag as provided by the CNP replication lag metric filtered by the Postgres Instance ($instance) pod:

cnp_pg_replication_lag{container="postgres", pod="$instance"}

replication_lag_example

Connections used

To calculate the connections used as a percentage, you can leverage the cnp_backends_total and cnp_pg_settings_setting metrics. This query calculates the percentage of connections used filtered by the Postgres Instance ($instance) pod:

100* (sum by (pod)(cnp_backends_total{pod="$instance"}) / max by (pod)(max_over_time(cnp_pg_settings_setting{name="max_connections",pod="$instance"}[30m])))

connections_example

Transaction ID age

This query shows the database transaction ID age as provided by the CNP database xid metric filtered by the Postgres Instance ($instance) pod:

(max by (pod) (cnp_pg_database_xid_age{container="postgres", pod="$instance"}))

xid_age_example

The age is relative to the workloads being run on your database and the following settings. See the Postgres documentation for details:

autovacuum_freeze_max_age. The default is 200000000.
vacuum_multixact_failsafe_age. The default is 1600000000.