Cluster health
The Status column on the Clusters page gives an overview of each PostgreSQL cluster's state. Autobase periodically queries the Patroni API and compares the reported cluster members with the expected number of PostgreSQL nodes.
Example of a three-node cluster with the healthy status:
Health statuses
| Status | Meaning |
|---|---|
healthy | Patroni reports at least the expected number of PostgreSQL members, and at least that many members are in the running or streaming state. |
unhealthy | Patroni reports the expected number of members or more, but fewer than the expected number are running or streaming. For example, a member may be present but stopped or still starting. |
degraded | Patroni reports fewer PostgreSQL members than expected. One or more database nodes are missing from the reported topology. |
unavailable | The platform previously connected to Patroni but can no longer retrieve cluster information from any known node's Patroni API. This can indicate an unreachable cluster or a connectivity problem between the platform and its nodes. |
maintenance | The interface temporarily displays this status while it tracks an active management operation for the cluster, such as applying configuration changes, scaling, or upgrading. Check Operations for progress and logs. |
A health status summarizes Patroni membership and member states. It does not evaluate application response times, CPU or disk utilization, replication lag thresholds, or every aspect of database availability. Use the dashboards to investigate performance and resource usage.
The maintenance status is an operation indicator in the current UI session. It does not mean that Patroni automatic failover has been paused. Patroni maintenance mode is a separate setting; see System Commands.
Deployment statuses
During initial deployment, you may also see:
| Status | Meaning |
|---|---|
deploying | Cluster deployment is in progress. |
ready | The deployment operation completed successfully. Subsequent Patroni health checks determine the cluster's health status. |
failed | The deployment failed. Open Operations to inspect the operation and its logs. |
Investigate a status change
Open the cluster's Overview page to review individual node roles, states, replication lag, and pending restart indicators. If a management operation is running or has failed, review Operations. Use Monitoring to correlate the change with system load, PostgreSQL activity, or replication behavior.
If the status is unavailable, check connectivity from the platform to the database nodes and the availability of their Patroni APIs.