Reading exceptions

The Exceptions page groups stack traces by fingerprint — two crashes with the same shape collapse into one row, with an occurrence count and timestamps.

Why grouping

If a single bug causes a hundred thousand stack traces, you don't want a hundred thousand rows. Grouping reduces that to one entry showing the trend.

The fingerprint is computed from the exception type + the top frames of the stack. Different exception types or different top frames are different groups.

  • Title: Exceptions · Auto-detected errors and exceptions grouped by fingerprint.
  • Refresh button top-right.
  • Source switcher in the left rail scopes the page.

Time picker

Presets: 15m | 1h | 6h | 24h | 7d plus explicit From/To datetime inputs.

Filter tabs

Five lifecycle filters across the top:

  • All — every fingerprint group, regardless of state.
  • Unresolved (default) — actively firing groups.
  • Investigating — someone has acknowledged and is looking.
  • Resolved — manually marked done. Will start a fresh group if it occurs again.
  • Ignored — muted. Does not count toward dashboards or alerts.

Sort by

Dropdown top-right. Options include Last Seen (default), occurrence count, first seen, etc.

Results table

Per-row columns:

ColumnMeans
(checkbox)Bulk-select for batch actions.
TYPE / MESSAGEException class + brief message (e.g., NullPointerException at PaymentService.charge:142).
COUNTOccurrences in the time range.
FIRST SEENWhen this fingerprint was first seen — ever, not just in the time range.
LAST SEENMost recent occurrence.
SERVICESService names that produced this fingerprint.
ACTIONSPer-row icons (mark investigating, resolve, ignore).
STATUSCurrent lifecycle state pill.

When no fingerprints match the filter:

> "No exceptions found for this filter"

Detail view

Click a row to open the fingerprint detail (separate page). Shows the full stack trace with most-recent-first ordering, occurrence chart over time, sample log line that produced the exception, tags from the originating log (service, environment, version, user_id when present), and any alert rules firing on this fingerprint.

Actions

Each fingerprint can be:

  • Mark Investigating — moves to the Investigating tab; useful for "I'm looking into this".
  • Resolve — moves to Resolved. New occurrences after resolve open a fresh group.
  • Ignore — moves to Ignored. Stops counting toward dashboards or alerts. Use for known-noisy errors you can't fix immediately.

Bulk-select multiple rows with the checkboxes to apply the same action to all.

What this is not

  • Not log search. It's specifically structured exception/stack data. Use Log Explorer for free-text log search.
  • Not bug tracking. Resolving here doesn't close a Jira issue. Wire up a webhook channel to push to your tracker.