View PostgreSQL Cluster Metrics

You can monitor the performance of your PostgreSQL v2 clusters by opening the metrics dashboard in Grafana directly from the DCD.

Prerequisites: An active PostgreSQL v2 cluster and access to the Grafana dashboard. For more information, see Create a PostgreSQL Cluster.

1

Navigate to PostgreSQL clusters

1. In the DCD, go to Menu > Databases > PostgreSQL.

A list of all PostgreSQL clusters is displayed.

2. Select a cluster from the list by clicking on its name. Alternatively, click Details in the OPTIONS column.

Select a PostgreSQL cluster

2

Open Grafana from the cluster details

1. From the Cluster details page, under the Observability section, select Open in Grafana.

2. Log in to the Grafana dashboard using IONOS CLOUD Single Sign-On (SSO). For more information, see Access Metrics from the Platform.

3

Configure the data source

1. From the Grafana home page, on the left-hand menu, go to Connections > Data sources.

2. From the Data sources page, select IONOS Telemetry.

Grafana metrics Data sources

3. In the Settings tab, go to the Authentication section > HTTP headers.

4. Select Add header. You must add an authorization header with a bearer token. To do so, configure the following:

• Header: In this field, enter Authorization.
• Value: Obtain the bearer token value from the Token Manager section in the DCD and prefix Bearer before the Token Value and add it here. For more information, see Generate authentication token.

5. In the Other section, select the HTTP method as GET.

Grafana HTTP Method

6. Click Save & test, available at the end of the Settings page.

4

View PostgreSQL database metrics

1. From the Grafana home page, on the left-hand menu, go to Explore > Metrics.

2. Select Let's start! to start metrics exploration and choose IONOS CLOUD Telemetry as the Data source.

Data source metrics

3. In the Search metrics, type the keywords such as postgres to view metrics related to PostgreSQL.

PostgreSQL metrics

Result: The Grafana metrics explorer displays available PostgreSQL metrics. You can search and visualize metrics to monitor cluster health and performance. For a full list of available metrics, see Metrics for PostgreSQL.

Metrics for PostgreSQL

The following metrics are available to gauge the health and performance of PostgreSQL clusters.

Database health and connections

Metric	Description
pg_up	Indicates whether PostgreSQL is reachable. A value of 1 means the instance is up; a value of 0 means it is unreachable. Alert immediately if the value is 0.
pg_stat_activity_count	The number of active sessions grouped by state. States include active, idle, idle in transaction, and others. Monitor this metric to understand connection usage patterns.
pg_stat_database_numbackends	The number of active connections per database. Compare this value with pg_settings_max_connections and alert when usage reaches 80% of the limit.
pg_settings_max_connections	The configured value of the max_connections PostgreSQL setting. Use this together with pg_stat_database_numbackends to calculate connection utilisation.
pg_settings_superuser_reserved_connections	The configured value of superuser_reserved_connections. Subtract this from pg_settings_max_connections to determine the effective connection limit available to regular users.

Database size

Metric	Description
pg_database_size_bytes	The size of each database on disk, measured in bytes. Monitor this metric to track growth and alert when approaching capacity thresholds.

Transactions and errors

Metric	Description
pg_stat_database_xact_commit	The cumulative number of committed transactions per database. Use rate() in PromQL to calculate transactions per second (TPS).
pg_stat_database_xact_rollback	The cumulative number of rolled-back transactions per database. A high rollback rate indicates application errors or logic issues.
pg_stat_database_deadlocks	The cumulative number of deadlock events per database. This value should remain close to zero; any non-zero reading warrants investigation.
pg_stat_database_conflicts	The cumulative number of query conflicts per database. This metric is most relevant when monitoring standby instances, where conflicts arise from replication activity.

Background maintenance and checkpoints

Metric	Description
pg_stat_bgwriter_checkpoints_timed	The cumulative number of checkpoints triggered by the scheduled checkpoint interval. This represents a stable baseline. Compare with pg_stat_bgwriter_checkpoints_req to assess checkpoint pressure.
pg_stat_bgwriter_checkpoints_req	The cumulative number of checkpoints triggered on demand because WAL volume exceeded the configured threshold. If this value consistently exceeds pg_stat_bgwriter_checkpoints_timed, consider increasing max_wal_size.
pg_wal_archive_rate	The rate at which WAL segments are archived. Monitor this metric to detect archiving backlogs or failures that could affect recovery point objectives.

Replication

Metric	Description
pg_replication_lag	The replication lag on a standby instance, measured in seconds. For critical workloads, alert when this value exceeds 30 seconds.
pg_replication_slots_active	The number of active replication slots. Inactive slots that retain WAL segments can cause disk exhaustion if not monitored and removed.

Vacuum and table activity

Metric	Description
pg_stat_user_tables_n_dead_tup	The number of dead (not yet vacuumed) tuples per table. High values indicate a vacuum backlog that may degrade query performance.
pg_stat_user_tables_last_autovacuum	The Unix timestamp of the last autovacuum run per table. Monitor the gap between runs; a long gap indicates that autovacuum is not keeping up with table activity.
pg_stat_user_tables_n_tup_ins	The cumulative number of rows inserted per table. Use rate() to calculate insert throughput.
pg_stat_user_tables_n_tup_upd	The cumulative number of rows updated per table. Includes both HOT and non-HOT updates.
pg_stat_user_tables_n_tup_del	The cumulative number of rows deleted per table. Use rate() to calculate delete throughput.
pg_stat_user_tables_n_tup_hot_upd	The cumulative number of HOT (Heap Only Tuple) updates per table. A high ratio of HOT updates to total updates indicates an effective fillfactor configuration.

Memory performance and query statistics

Metric	Description
pg_statio_user_tables_heap_blks_hit	The cumulative number of in-memory buffer hits for table data. Use together with pg_statio_user_tables_heap_blks_read to compute the hit ratio. The target hit ratio is above 99%.
pg_statio_user_tables_heap_blks_read	The cumulative number of disk reads for table data. A high value relative to in-memory hits indicates memory misses; consider increasing shared_buffers.
pg_stat_statements_sum_querycount	The total number of query executions across all tracked statements. Requires the pg_stat_statements extension to be enabled in the PostgreSQL configuration.