Sentry
Open an investigation from a Sentry issue alert and read the issue, its event counts and its recent events.
Read the issue and its events
A Sentry issue alert opens an investigation, which reads that issue through the connection: its short ID, level, status and first and last seen times. It counts the issue’s events in one-minute buckets from 30 minutes before the alerting event to 30 minutes after it, and reads up to 25 events in that window.
From each event, Proofline keeps the title, the outermost exception’s type and value, the level, release and environment tags, and the module and function names of up to 32 in-app frames. It drops request and user data, breadcrumbs, local variables, contexts and other tags, then checks the kept text for protected values and masks credentials. A read stops at one day, 8 MiB of responses or 30 seconds.
Investigate a new error after a release
Suppose a Sentry issue starts alerting soon after a deploy. The per-minute event counts sit beside the environment’s deploys and merges in the timeline, and each event’s release tag shows the release Sentry recorded for it.
Event text is untrusted data. A read establishes neither continuous coverage, the absence of an error nor service health, and Proofline records no close for a Sentry alert.
Connect Sentry
An organization administrator creates a Sentry internal integration with exactly the org:read, project:read and event:read scopes, then connects the Sentry organization with its data region, slug, token and client secret. Link exactly one environment to the connection: an alert through a connection linked to none or several opens no investigation. Self-hosted Sentry is not supported, and alarm investigations are not available to new workspaces today. Talk to the team about availability.