Is there an existing issue for this?
How do you use Sentry?
Sentry Saas (sentry.io)
Which SDK are you using?
@sentry/nextjs
SDK Version
10.73.0
Framework Version
Next 16.3.4
Link to Sentry event
No response
Reproduction Example/SDK Setup
Any app that adds more than maxBreadcrumbs (default 100) breadcrumbs to a scope.
Steps to Reproduce
- Keep a scope alive long enough to add 101 breadcrumbs (a long-lived server scope, or a browser page with console/fetch/click breadcrumbs).
- Look at the client reports the SDK sends, or at Sentry → Stats → Outcomes for the project.
Expected Result
Breadcrumb eviction from the ring buffer is not a discarded log item. Either it is not reported as a client report at all (a breadcrumb is context, and 100 is the intended cap), or it is reported under a breadcrumb-shaped reason/category rather than log_item.
Actual Result
Scope.addBreadcrumb records a client report with reason buffer_overflow and category log_item every time the breadcrumb buffer evicts an entry:
// packages/core/src/scope.ts (develop), same in @sentry/core 10.73.0 build/cjs/scope.js:394
this._breadcrumbs.push(mergedBreadcrumb);
if (this._breadcrumbs.length > maxCrumbs) {
this._breadcrumbs = this._breadcrumbs.slice(-maxCrumbs);
this._client?.recordDroppedEvent('buffer_overflow', 'log_item');
}In Sentry's Stats page this shows up as discarded log items, client_discard · buffer_overflow, and in our project it dwarfs accepted logs: 84,981,094 "discarded" against 1,692,344 accepted log items over 7 days. We spent an investigation working out that none of those were log lines: the logs buffer flushes at 100 items / 800 KB / 5 s and never discards for this reason, so the only source of buffer_overflow · log_item in the SDK is breadcrumb eviction. The mislabel makes the Stats page read as if 98% of logs are lost in-process, when nothing is.
Suggested fix: drop the recordDroppedEvent call in addBreadcrumb, or give it its own reason/category so it cannot be read as log loss.
Is there an existing issue for this?
How do you use Sentry?
Sentry Saas (sentry.io)
Which SDK are you using?
@sentry/nextjs
SDK Version
10.73.0
Framework Version
Next 16.3.4
Link to Sentry event
No response
Reproduction Example/SDK Setup
Any app that adds more than
maxBreadcrumbs(default 100) breadcrumbs to a scope.Steps to Reproduce
Expected Result
Breadcrumb eviction from the ring buffer is not a discarded log item. Either it is not reported as a client report at all (a breadcrumb is context, and 100 is the intended cap), or it is reported under a breadcrumb-shaped reason/category rather than
log_item.Actual Result
Scope.addBreadcrumbrecords a client report with reasonbuffer_overflowand categorylog_itemevery time the breadcrumb buffer evicts an entry:In Sentry's Stats page this shows up as discarded log items,
client_discard · buffer_overflow, and in our project it dwarfs accepted logs: 84,981,094 "discarded" against 1,692,344 accepted log items over 7 days. We spent an investigation working out that none of those were log lines: the logs buffer flushes at 100 items / 800 KB / 5 s and never discards for this reason, so the only source ofbuffer_overflow · log_itemin the SDK is breadcrumb eviction. The mislabel makes the Stats page read as if 98% of logs are lost in-process, when nothing is.Suggested fix: drop the
recordDroppedEventcall inaddBreadcrumb, or give it its own reason/category so it cannot be read as log loss.