Throughput per second - eps() or epm()

[jorgeavilacardenosa]

Environment

SaaS (https://sentry.io/)

Steps to Reproduce

Hi! I’m trying to build a custom dashboard to understand Throughput per second, grouped by endpoint.

I’m using the eps() or epm() functions, but the values I’m getting don’t seem accurate. I’m currently sampling 1% of the events, and even after multiplying by 100, the data still doesn't look right.

I'm currently sending about 40 requests per second to the system, and this is the data I see:

The application that i'm monitoring is a python app with this initialization:

sentry_sdk.init( dsn=SENTRY_DSN, release=SENTRY_RELEASE, traces_sample_rate=0.01, profiles_sample_rate=0.01, environment=ENV, integrations=[ StarletteIntegration( transaction_style="endpoint", failed_request_status_codes={401, *range(500, 599)}, ), FastApiIntegration( transaction_style="endpoint", failed_request_status_codes={401, *range(500, 599)}, ), ], )

What am I doing wrong? Thanks

Expected Result

I'm expecting to see around 30-40 requests per second.

Actual Result

Same issue with the events per minute funciton

Product Area

Dashboards

Link

https://monkeytaps.sentry.io/

DSN

No response

Version

No response

[getsantry]

Assigning to @getsentry/support for routing ⏲️

[getsantry]

Routing to @getsentry/product-owners-dashboards for triage ⏲️

[gggritso]

@jorgeavilacardenosa thanks for raising this! We believe it's a bug with formulas involving eps(), we're taking a look