Errors
The Errors page is a focused, first-class view of everything that went wrong in your app — failed executions and warning/error logs — so you can spot and fix problems fast.
Found under “Observe → Errors”
In the project sidebar, this page lives in the Observe group next to Logs and Usage. It's distinct from the Logs (Console) stream: Logs show everything in real time, while Errors surface only the failures, already filtered and counted for you.
Overview
Errors collects two kinds of problems into one table:
- Failed executions — endpoint or workflow runs that errored out or timed out.
- Warning / error logs —
WARNandERRORlog entries emitted while your app ran.
Each item shows the time, type, environment, endpoint, message, and how long the run took. Recent items are kept for 7 days.
At a glance
The page header shows four counters so you can gauge the situation before digging in:
| Counter | What it counts |
|---|---|
| Total | All items currently matching your filters |
| Failed executions | Endpoint/workflow runs that failed or timed out |
| Errors | Items at the error level |
| Warnings | Items at the warn level |
Filtering
Combine any of the filters below to narrow down what you're chasing:
| Filter | Options |
|---|---|
| Time range | Last hour, last 24h, last 7 days |
| Environment | All, Live, Draft |
| Level | All, Errors, Warnings |
| Endpoint | Filter to a single endpoint (e.g. create-order) |
| Search message | Match any substring of the error message |
Use the Environment filter to separate problems in your published app (Live) from ones you hit while iterating in Draft.
Reading the table
Each row describes one failure:
| Column | Shows |
|---|---|
| Time | When the failure happened |
| Type | failed or timeout for executions; error or warn for logs |
| Env | Live or Draft |
| Endpoint | The endpoint that ran (when applicable) |
| Message | The error message, plus its log category |
| Duration | How long the run took before failing |
| Action | Open the full execution Trace when one is available |
Inspecting a trace
When a failed execution has a recorded trace, the row's Trace button opens a step-by-step view of that run — which node ran, what it received, and where it broke. This is the fastest way to find the exact step that failed.
Your AI can pull the same data: the cloud MCP dypai_trace tool returns an execution trace so the agent can diagnose and fix the failing step for you.
The list shows up to the 200 most recent items in the selected window. Hit Refresh to pull the latest, or narrow the filters if you're getting too many results.