Errors dashboard & triage
Fingerprint-grouped error issues with a status workflow, occurrence timeseries, and automatic re-open on regression.
The Errors page (Errors in the sidebar, /dashboard/<projectId>/errors) is where the errors reported by see() land. Each distinct issue is grouped by fingerprint — so a thousand occurrences of the same problem are one row, not a thousand — and shown with its consequence sentence, occurrence trend, and triage state.
This page displays errors. Producing them is the SDK's job — wrap a catch with
see(err).causes_the(…).to(…). See Error reporting for the producing
side.
What a row shows
| Column | What it is |
|---|---|
| Error | The consequence sentence (e.g. "TypeError causes the checkout to use cached prices") with the raw message underneath. |
| Trend · 14d | A daily occurrence sparkline, the total count, and the % change of the last 7 days vs the previous 7. Errors are lower-is-better, so an upward delta renders red. |
| Impacted | Distinct users affected in the last 14 days. |
| First seen / Last seen | When the issue first and most recently fired. |
A project-wide Error rate · 24h banner sits above the table. The list polls every 30 seconds, so the page stays close to live without sockets.
The status workflow
Every issue is in one of three states, shown as the tabs Open, Resolved, Ignored (plus All):
open ──resolve──▶ resolved
open ──ignore──▶ ignored
resolved/ignored ──reopen──▶ open- Open — actively counts toward your open total.
- Resolved — you've fixed it. It moves out of the open list but stays grouped.
- Ignored — noise you've chosen to mute; hidden from open and no longer counted.
Flip status from the per-row ⋯ menu (Mark resolved, Ignore, Reopen), from the detail panel, or in bulk by selecting rows and using the bulk-action bar.
From error to ticket
A grouped issue doesn't stay a passive row. Once an issue passes 10 occurrences, Shipeasy files a ticket for it into the project's ops queue, so it can be picked up, assigned, and fixed like any other bug. If an issue then goes quiet, it auto-closes after 3 days without a new occurrence. Both numbers are per-project settings — change them under Settings → Ops ("Create ticket after N occurrences" and "Auto-close resolved errors").
Because issues are grouped by fingerprint, a recurrence of something you marked resolved re-opens the existing issue instead of filing a brand-new one — so the history (first-seen, full trend) stays intact and you're not chasing duplicates.
Caused-by chains
When code does see(err) and re-throws, the outer report links to the inner one as a caused-by chain rather than double-counting. The detail panel surfaces the upstream issue, so you can trace a wrapped error back to its root cause without it inflating your counts.
Error lifecycle
Two numbers decide when an error closes itself and when it becomes a ticket. Both live under Settings → Ops → Error lifecycle:
| Setting | Default | What it does |
|---|---|---|
| Auto-close quiet errors | 3 days | Resolves an open error once it has gone unseen this long. It reopens by itself when it happens again. 0 turns auto-close off. |
| Open a ticket after | 10 occurrences | How many times an error has to happen before it is filed into the queue on Home as a ticket. Tuning it needs a paid plan; every plan files at the default. |
Filing an issue from an error
Each row has a File an issue action. It creates an error-type ticket on Home, the project's ops queue, pre-filled from the tracked error and linked back to it — once filed, the row shows a #42 chip instead of the file button. From there a connector can mirror it to GitHub. Worker-side auto-filing uses the same link, so manually-filed and auto-filed tickets de-duplicate.
Error reporting with see()
Report every handled exception with see() so it folds into the errors primitive with a product consequence.
Inbound alerts
File incidents from Google Cloud Monitoring, Prometheus Alertmanager, Grafana, AWS CloudWatch, Azure Monitor, Datadog — or any tool that can POST JSON — as alert tickets in your ops queue, with no relay to build or keep alive.