Dashboards

Dashboards are your own views across every dataset Errora stores: a line chart of error volume next to p95 latency next to a table of your slowest transactions.

Create one

Dashboards → New dashboard, name it, then add widgets. Dashboards belong to the organization, so everyone with access sees the same board.

Widget anatomy

Each widget is: a dataset, a visualization, one or more aggregates, an optional group-by, and optional filters.

Visualizations

line, area, bar, table, big_number.

Datasets and what they can answer

Dataset	Aggregates	Group by / filter
errors	count(), count_unique(issue)	level, environment, release, platform, server_name
transactions	count(), avg(duration), sum(duration), failure_rate(), p50–p99``(duration)	transaction, op, status, environment, release, platform
logs	count()	level, environment, release
metrics	the metric's own aggregates (sum, avg, min, max, percentiles, count_unique)	any tag the metric carries
profiles	count(), avg(duration), sum(samples), p50–p99​(duration)	transaction, environment, release, platform
issues	count(), sum(events), sum(users)	issue attributes

The builder reads this catalog from the server (GET /dashboards/catalog), so it can only offer queries the backend can actually answer — and a widget you saved last month still renders, because the serializer validates against the same registry.

Example widgets

Question	Dataset	Visualization	Aggregate	Group by
Are we getting worse?	errors	line	count()	level
Which endpoint is slow?	transactions	table	p95(duration)	transaction
How healthy is checkout?	transactions	big number	failure_rate()	— (filter transaction:/checkout)
Log volume by severity	logs	area	count()	level
Orders today	metrics	big number	sum of orders.created	—
Slowest profiled routes	profiles	bar	p99(duration)	transaction

Arranging and sharing

Widgets can be reordered (drag, or PATCH .../widgets/reorder). Everyone in the organization with project:read can view the board; editing needs the usual write capabilities. A dashboard is scoped to the organization, and each widget resolves against the projects you can see.

API

GET    /api/v1/dashboards/catalog
GET    /api/v1/organizations/{org_id}/dashboards
POST   /api/v1/organizations/{org_id}/dashboards
GET    /api/v1/organizations/{org_id}/dashboards/{id}
PATCH  /api/v1/organizations/{org_id}/dashboards/{id}
DELETE /api/v1/organizations/{org_id}/dashboards/{id}
POST   /api/v1/organizations/{org_id}/dashboards/{id}/widgets
PATCH  /api/v1/organizations/{org_id}/dashboards/{id}/widgets/reorder
PATCH  /api/v1/organizations/{org_id}/dashboards/{id}/widgets/{widget_id}
DELETE /api/v1/organizations/{org_id}/dashboards/{id}/widgets/{widget_id}
GET    /api/v1/organizations/{org_id}/dashboards/{id}/widgets/{widget_id}/data

Troubleshooting

Symptom	Cause / fix
An aggregate isn't offered	It isn't in that dataset's catalog entry — percentiles exist for durations, not for counts
Widget is empty	No data in the period, or a filter value that never occurs (check spelling of environment/release)
Percentile looks off	Percentiles are estimated from a bounded sample of the underlying values
Group-by column missing	Only the catalog's declared columns are selectable; anything else would let user input reach the ORM as a field name