#429 Rate limits exceeded.... but not understanding where

1 messages · Page 1 of 1 (latest)

jaunty latch
#

In our local, staging, and prod environments we are experiencing an outage where we keep getting rate limits exceeded but can't track where in the dashboard or why. When putting a breakpoint in localhost and testing with a 3000 token prompt, we are getting this issue consistently. We don't have enough users entitled for our AI to hit the rate limits and are treating this as an S1 outage on our part and are having issues tracking the API keys which (could) be spamming the API (yet our bill and usage doesn't reflect this)

image.png image.png
grand parcel
#

the dashboard is likely showing a different org than the one being used for requests

jaunty latch
#

@grand parcel I checked the orgs out of sanity and they are the same. Although- I noticed that when I’m logged in with one account, I can’t see all the api keys but when I log in with another account in the same org, we see them. Super confused

grand parcel
#

im not sur eof what you mean, log i nwith the account that own the org and you will be able to see the usage

jaunty latch
#

@grand parcel I have two accounts that are set as owners in the same org (same one in the error). Only one of the two accounts can see the api keys despite having the same rbac role owner.

And then on top of this, we can’t figure out what’s causing the rate limit to trigger 😅. Why I think it’s a genuine bug

grand parcel
#

this is indeed very weird, if yo uchecked everything, I can only assume it is a bug

#

that one might require talking to the support

jaunty latch
#

@grand parcel suddenly its working this morning and not sure why, still not showing the usage -- even the last usage of the same key is showing nothing

image.png image.png image.png
#

Does anyone from OpenAI have an RCA on this I can share down stream?

grand parcel
#

nice

jaunty latch
#

@grand parcel not to harp on this but is there a way to figure out how this happened? I appreciate you being responsive 😊 the support from the intercom chat is not responding

grand parcel
#

you can still contact the support to ask if it was a bug on the side of OpenAI