The landing screen at /_ answers "is the server healthy, and what has it
been doing?" Everything on it is computed from the server's own request log and
job history — there is no external metrics service to configure.
The numbers#
Four figures sit across the top, all over a rolling 24-hour window:
| Tile | What it counts |
|---|---|
| Requests (24h) | Total requests handled |
| Error Rate | Share of those requests that failed |
| Active Sessions | Sessions currently valid |
| Tables | Tables in the database |
Underneath, Requests over time breaks the same 24 hours into hourly buckets, so a traffic spike or an outage is visible as a shape rather than a single number.
Exclude admin filters your own dashboard traffic out of the figures. Leave it on when you are trying to read real user activity — otherwise browsing tables inflates the request count you are looking at.
Top Errors (24h)#
The failures from the same window, grouped so the repeated ones rise to the top rather than scrolling past one at a time. When nothing has failed, the panel says so — No errors in the last 24h is a useful thing to be able to confirm at a glance.
For the full log rather than the summary, use zonai db.
Cron Jobs#
Every job defined in your project is listed with its most recent outcome:
| State | Meaning |
|---|---|
| Succeeded | Last run completed |
| Failed | Last run threw |
| Running | Currently executing |
| Never run | Defined, but not yet triggered |
Run triggers a job immediately, without waiting for its schedule. That is the fastest way to check a job you have just written, and it is the same manual trigger described in Running Manually — so the usual caution applies: the job does real work against real data.
A refresh control re-reads job state without reloading the page, which is useful while watching a long job finish.
A job that mutates rows will also push updates to any open live queries watching those rows — including a table open in another dashboard tab.
Related#
- Cron Jobs Overview — how scheduled jobs are compiled and run
- Catch-Up Logic — what happens to runs missed while the server was down
- Browsing & Editing Data — inspecting the rows a job touched