Dashboards

For SPG99, it is useful to maintain at least several dashboards or logical views.

1. Database lifecycle

Useful indicators:

• the current state;
• cold-start frequency;
• time to ready;
• the number of databases in error;
• the distribution of ready / stopped;
• scale_state across databases.

2. Writer autoscaler

This is a new important view. It is useful to see:

• current_profile and target_profile;
• the number of databases in PREPARING, DRAINING, COOLDOWN, and FAILED;
• freeze/drain/cutover duration;
• reasons for handoff failures;
• databases where pinned or session-heavy traffic is delaying cutover.

3. Gateway and client entry

Useful things to see:

• active connections;
• lease activity;
• checkout wait / timeout;
• TLS errors;
• signs of pooling overload.

4. Compute and PostgreSQL

Here you usually watch:

• CPU;
• memory;
• connections;
• long-running queries;
• signs of degradation or overload.

5. The storage chain

It is important to see the state of Pageserver and Safekeeper, especially if the problem looks like:

• prolonged bootstrap;
• slow cold start;
• WAL quorum degradation;
• lag while restoring startup state.

Practical meaning

A good set of dashboards helps you quickly understand whether the issue is local to one database or systemic for the platform. After the rollout of soft basebackup and the writer autoscaler, this is especially important: the same user-facing symptom can now mean cold start, handoff, pinned workload, or storage lag.