[Feature Request] Rate limit details

[jeengbe]

Problem Statement

We have a rate limit configured.

Do projects exceed it? Which ones do? How close are we?

Can also lead to issues like #44833

Solution Brainstorm

More information on the rate limit page.

[getsantry]

Assigning to @getsentry/support for routing, due by Tuesday, February 21st at 09:28 (yyz). ⏲️

[jeengbe]

Also, (slightly out of context) what's the point of the rate limit option? Shouldn't this be covered by spike protection entirely?

[getsantry]

Routing to @getsentry/enterprise for triage, due by Wednesday, February 22nd at 5:00 pm (sfo). ⏲️

[Dhrumil-Sentry]

@jeengbe Thanks for bringing up this issue. You can find more information about Rate Limiting here

Spike Protection protects your quota from being rapidly depleted. Still, if you see a sustained high volume of events over several days then the Spike protection algorithm will start adjusting since it interprets your event counts as the "new normal" and accordingly spike protection thresholds will start trending higher.

Rate limit is an explicit way for customers to say drop any events if Sentry sees 200+ errors in 1min- Rate Limits are rigid limits and if used aggressively can lead to you missing critical errors and issues

[jeengbe]

Ah I see, thanks @Dhrumil-Sentry! Sort of like a safeguard on top of spike protection.

This should say something related with the rate limit instead of "Dropped". I had no idea we drop 60% of errors due to rate limiting.

[Dhrumil-Sentry]

@jeengbe Rate limits are available for Business Plans only and your Org does not have any rate limits, Dropped errors are errors dropped due to invalid data, rate limits, quota limits, or spike protection.

It's likely that your Org is receiving more errors than your monthly quota leading to Sentry dropping those errors

But I agree we should have a better breakdown of dropped errors and I am adding this to our backlog

[jeengbe]

Hi @Dhrumil-Sentry

Then there is probably something wrong with your rate limiting stuff. Not only can I use the UI without restrictions, error events are capped at precisely 2.5k an hour, which corresponds with our set rate limit.

Of all things, rate limits are no feature to lock behind a "Business plan", as users who run the Team plan most likely care about cost with rate limits being a crucial last barrier if things go really south.

[getsantry]

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