How does Grafana help you investigate failed bookings?

Grafana can plot how often bookings fail, then help you inspect the logs you collected during the same time window. This profile covers Grafana OSS, the open-source dashboard and alerting software.

Measurements help you see when the problem started and how often it happened. Logs are notes a service recorded while it was running. A trace follows a single request between connected services. Grafana retrieves these records from the systems that hold them. For example, you might connect Prometheus for measurements, Loki for logs and Tempo for traces.

Grafana Cloud is a separate service run by Grafana Labs. It can provide managed storage and other services under its own plans. An installation of Grafana OSS still needs somewhere to collect and keep your app’s signals.

What must be connected before the dashboard helps?

Start with one data source that counts booking attempts and failures. A data source is the connection Grafana uses to ask the system that holds those records for specific results.

Match the time window and the app or environment in both counts. Check the connection with one known record. An empty panel might mean the wrong time range, the wrong source or a missing permission. Make sure you can see an ordinary test request before taking an empty error chart as good news.

Keep the connection limited to the records the intended users can see. By default, a Grafana organization’s viewers can query its connected data sources beyond the questions on a dashboard. Use restricted access at the source and keep private records out of any shared practice connection. Fine-grained Grafana data source permissions are an Enterprise or Cloud feature.

Once the chart gives a reliable signal, you can use an alert rule to notice when a supported source meets a condition you choose. Check what a quiet or missing-data period should look like before adding a notification destination.

How can you tell whether the spike is worth investigating?

Use these fictional counts on paper. Compare the share of bookings that failed before deciding whether an increase in the error count deserves a closer look.

Choose the first window to inspect
Five times as many attempts brought five times as many failures in the second window, so the share stayed the same. Start with the third window, where 25 of 100 attempts failed.
Decide what to open next
Look for booking service log notes in the third window, keeping the same environment selected. If you had recorded traces, pick one trace for a failed booking to see which steps it reached.
Interpret an empty result
Imagine the chart has data but the log panel is empty. First check whether the booking service writes to that log source and whether you chose the matching time range. Write down which evidence you would need before deciding why bookings failed.

Two fictional windows for the same booking service
Time window	Attempts	Failures	Share that failed
10:00–10:05	20	2	10%
10:05–10:10	100	10	10%
10:10–10:15	100	25	25%

On a small screen, scroll the table sideways to read every column.

What does Grafana OSS cost to run, and what can you move?

Grafana OSS uses the AGPLv3 open-source license. For a self-managed setup, budget for the machine running Grafana, the systems storing its data and whoever keeps them working.

To move a dashboard, use Export as code to save its layout and queries as a JSON file. Set up the intended data sources on the receiving instance and open a known time window to confirm the panels still find their records. You can export the selected panel results as a CSV where the panel supports it. A dashboard definition or panel download covers a different part of your setup than the stored history.

For a recovery plan, keep the Grafana configuration, plugins and database as documented for your installation. Decide separately how you will retain or restore the underlying measurements, logs and traces.