PostgreSQL Dashboard
Use the Dashboard to see what one PostgreSQL instance, or one of its databases, does: its active backends, their waits and the busiest statements.
Before you begin
- You need the View PostgreSQL and View Dashboard permissions. Open the Dashboard from the Databases page or the database bar.
- Scops keeps the activity samples for 15 days by default (see Retention).
Show one database
In the database bar, click All databases next to the name of the instance.
Select a database. If the list is long, type part of the name in Find a database.
The page shows only that database. The other pages of the instance keep your choice. To show the full instance again, select All databases.
Read the values at the top
The four values show the newest sample. A value has a color only when it needs attention. Click a value to open its history.
| Value | What it shows |
|---|---|
| Active sessions | The sessions that work now. |
| Blocked | The sessions that wait on a lock that a different session holds. |
| Connections | The client sessions against max_connections, for example 41 of 100, and the sessions idle in a transaction. Amber from 80%, red from 90%. |
| Oldest transaction | How long the oldest transaction is open, and its pid. None means that no transaction is open. Amber from one hour. |
Connections always shows the full instance. When you show one database, it has the tag instance. If there is no sample, the values show a dash, and a line next to them tells you why.
Open the session history
Click a value at the top. The Sessions window opens on its tab.
Select a period: 1H, 1D, 1W or 1M.
To go to the related page, click Open Live Activity or Open Metrics.
| Tab | Chart | Table under the chart |
|---|---|---|
| Active sessions | Active backends: the sessions that work, with their parallel workers, on average. | None. |
| Blocked | Lock waits: the backends that wait on a lock, on average. | Blocking sessions: the session at the root of each chain, and how many sessions wait behind it. |
| Connections | The most client sessions at one time, the sessions idle in a transaction, and max_connections. | Client sessions by state. |
| Oldest transaction | The longest open transaction, in minutes or hours. | Open transactions, the oldest first. |
The tables show the last minute, with 8 rows at most.
Open the findings
Click Findings next to the values. The number shows the open warning and critical findings. It is red when one of them is critical.
In the Findings panel, click a row to read the finding.
0 means that no warning or critical finding is open. A dash means that Radar Engine does not watch this instance. Alert log, in the footer of the panel, opens all alerts and findings of the instance on Incidents.
Choose the time window
| Button | Window |
|---|---|
| Live | The last hour. The page refreshes every 15 seconds. |
| 1D, 1W, 1M | 24 hours, 7 days or 30 days, in hourly, 4-hour or daily bars. The window moves every 5 minutes. |
| Calendar button | Any range. Select From and to, then click Apply. |
To move the window, drag the box on the strip under the chart. To change its length, drag an edge of the box. A window is one hour at the least. All times use the clock of the instance.
Read the Activity chart
The chart stacks the average active backends: the sessions that work, with their parallel workers. To change the split, select Wait Event Type (the default), Wait Event, Database, User, Application or Client in the chart header. If a choice does not show, look under More.
| On the chart | Meaning |
|---|---|
| CPU | Backends that run and do not wait. The Waiting On column calls this On CPU. |
| A dashed line, for example 8 CPUs | The CPU count of the host. It shows only when Scops reads the host over SSH (see PostgreSQL permissions). |
| Usual | What the instance or the database usually does at this time of day. It shows after three days of samples. |
| A gap | Scops could not sample. A gap is not zero. |
| A status shape | A Radar finding opened there. Click it to open the finding. |
In a window of one day or more, click a bar to make its time the window.
Select a range
Drag across the chart, or click one point. A chip in the chart header shows the range.
To clear the range, click the chip.
The Breakdown and the Statements then show only that range. Without a range, they show the full window, or the last 5 minutes in Live.
Filter by wait, database, user or client
In Breakdown, select Wait Event, Database, User, Application or Client.
Click a row. The chart, the Breakdown and the Statements show only that value.
To show all again, click the row again, or the chip in the chart header.
When you show one database, the chart and the Breakdown do not offer Database.
Find the busiest statements
The Statements tab lists the statements of the window or of the selected range, the busiest first. Scops identifies a statement by its query_id.
| Column | What it shows |
|---|---|
| Statement | The query_id. Work without a query_id shows its command. A tag such as ×4 means that it ran in parallel on up to 4 backends. |
| Activity | Its share of the activity, split by wait event type, and its average active backends. |
| Statement Text | The text, when Scops has it. |
| Waiting On | Its most frequent wait, or On CPU. |
| Database | Its database. Shows only with All databases. |
Open or find a statement
- Click a row. The statement details open for the same window. A row without a
query_iddoes not open. - To find statements by a piece of their text or a table name, type it in the box next to the tab. The list shows only the statements of the window that contain it.
- To open a statement by its
query_id, type it in the same box and press Enter. If the statement did not run in the window, first select its database in the database bar.
The box says Find a statement by text, table or query_id. To clear it, press Esc.
Related
Updated November 22, 2026