Skip to main content

Failed messages page (WebApp)

The Failed page (/Failed, under OBSERVE in the sidebar) is one place to find and act on every unresolved failure — events currently Failed, DeadLettered or Unsupported — across all the endpoints you can read. The sidebar badge shows the size of that backlog.

What the page shows​

  • Range — 1h, 12h, 1d (default), 3d, 7d or 30d. When failures older than the range exist, the page says how many and offers to widen it.
  • Summary tiles — totals per status in the range, the number of endpoints affected and the busiest bucket.
  • Failures over time — a stacked bar chart of failures per bucket (5 minutes for 1h up to a day for 30d), split by status or by endpoint. A failure counts in the bucket of the time it last entered its status (UpdatedAt), so the chart and the list always describe the same events. Click a bar to narrow the list to that window; click it again, or the chip, to clear it. The Metrics page's failure series is message history instead, and includes failures that were since resolved.
  • Filters — endpoint, event type, event ID, last message ID, session ID, publisher (From), subscriber (To), error text (case-insensitive substring) and status toggles.
  • Views
    • List — the endpoint page's columns plus Endpoint and the last error, with the "N blocked" chip for failures holding deferred messages in their session. Resubmit and Skip per row or for the selection. Columns can be shown or hidden.
    • By endpoint — counts per endpoint and status; Show failures narrows the list.
    • By error — the Insights grouping (error category, then normalized pattern) over every matching failure. Expand a category into its patterns and a pattern into its failures; Resubmit or Skip a whole category, a pattern or one failure. Up to 5,000 failures are grouped; the view says so when a filter matches more.

The endpoint page's Failed tile opens this page filtered to that endpoint. Filters live in the URL, so a view can be shared or bookmarked.

Access​

The page covers the endpoints you hold Reader on. Naming an endpoint you cannot read in the filter is refused (403) and audited. Every search and error grouping writes a SearchEvents audit row. Resubmit and Skip keep their own Contributor checks.

API​

POST /api/failed/search, POST /api/failed/histogram and POST /api/failed/error-groups — see WebApp REST API. They are backed by IMessageTrackingStore.GetFailedEventsAcrossEndpoints and GetFailedEventHistogram, which every storage provider implements (see Storage providers).