Check platform health and scheduled jobs
Use Platform health to see whether jobs, data sources, webhooks, lists and AI are working, and know what to do when something needs attention.
tocOn this page
Much of FlowSentry's work happens in the background: refreshing dashboards and registers, batching cash transaction reports, rescreening the book and delivering webhooks. Platform health gives administrators one page to check that all of it is running for your institution.
Before you start
- Platform health (
/ops/health) is for the Administrator role. Open it from the Admin module, under Platform operations on the Administration hub. - The Ops console, where scheduled jobs can be run on demand or disabled, belongs to the platform operator who runs the shared service. Tenant administrators do not have it. If a job needs to be rerun or paused, ask your FlowSentry service provider.
What Platform health shows
| Panel | What to look for |
|---|---|
| Scheduled jobs | Each job's latest run and any failures in the last 24 hours |
| Ingestion sources | Freshness and lag against each source's silence SLA, and when it was last seen |
| Webhooks | Deliveries pending, delivered and dead-lettered (given up after retries) |
| Platform sanctions and PEP lists | When each list was last fetched |
| AI usage | Calls, tokens and cost |
What job statuses mean
| Status | Meaning |
|---|---|
| Succeeded | The run finished normally |
| Failed | The run stopped with an error. The job is designed to leave data consistent and retry on its next run |
| Timed out | The run took longer than allowed and was stopped |
| Skipped | The run did not do any work this time; occasional skips are normal |
| Never run | The job has not run yet, for example just after go-live |
A single failure followed by a success usually needs no action. Repeated failures or time-outs on the same job do: contact your service provider and tell them which job and when.
Why a dashboard or register may lag
Dashboards and the customer register are built from summaries that background jobs refresh about every 15 minutes. They do not read live transaction data. So:
- a customer or transaction loaded a few minutes ago may not appear yet
- after a new institution goes live, figures fill within about 15 minutes.
If figures are still missing after half an hour, check the scheduled jobs panel for failures and then contact your service provider.
When a data source goes silent
Each source has a silence SLA: the number of minutes without data that counts as a problem. A source past it shows Needs attention.
- Check whether silence is expected. Some systems send little at night or over a public holiday.
- Open Data sources (
/integration/sources) and check that the source is still Live and not Suspended. - Open Ingestion runs (
/integration/runs) to see whether data is arriving but being quarantined. - Ask the team that runs the sending system whether it is up and still sending. Rejected signatures often mean the key was rotated on one side only.
- If the sending side looks fine, contact your service provider.
Warning: While a source is silent, transactions from that system are not being monitored. Tell your MLRO if a gap may affect monitoring coverage, and follow your institution's procedure for recording it.
Other things worth checking
- Webhooks dead-lettered. The receiving system refused or did not answer after repeated retries. Check the endpoint is reachable, then look at Recent deliveries in Tenant settings.
- Lists last fetched. Platform sanctions and PEP lists are kept by the FlowSentry list desk. If a list looks out of date, raise it with your service provider.
- FX rates. A missing rate in FX rates (
/admin/fx) means rules with thresholds in that currency cannot evaluate. Keep rates current. - AI usage. A sudden rise in calls or cost is worth a look in the AI usage dashboard.
Tip: Make checking Platform health part of your daily routine. A few seconds each morning catches most problems before analysts notice them.