Update max rate-limiting metadata #36734
Conversation
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎ 1 Ignored Deployment
|
octavia-squidington-iii
added
area/connectors
| @@ -16,7 +16,7 @@ data: | |||
| githubIssueLabel: source-sentry | |||
| icon: sentry.svg | |||
| license: MIT | |||
| maxSecondsBetweenMessages: 64800 | |||
| maxSecondsBetweenMessages: 60 | |||
There was a problem hiding this comment.
Checking the Sentry API doc, it seems like the rate limit window is a second: We restrict both how frequently a request is made (requests per second rate limit) <...>. What is the reasoning for having 60 seconds here?
There was a problem hiding this comment.
I was originally a bit confused by this as well, but when testing and viewing the response headers, the x-sentry-rate-limit-reset was being set to the next full minute from the request (so a request sent at 12:00:02 will return a reset time of 12:01:00, as will a request from 12:00:36). Worth noting that I ran some more tests and some of the streams do seem to have smaller windows, but the events stream in particular consistently sets this value to the next full minute.
There was a problem hiding this comment.
Can we document this somewhere as it would be interesting to know where this comes from. I'm also curious is HTTP requests to sentry can take more than a minute, maxSecondsBetweenMessages should be increased. I don't know how big the responses can be with big datasets but there is that to consider
There was a problem hiding this comment.
Sure thing, just added a comment for clarity 👍 I agree this value may actually be higher given it's not as set in stone as their docs would suggest, but my understanding is also that at the moment we don't actually do anything with this value unless it's at least 5400 seconds (or 90 minutes). Is that incorrect?
What
Updating max rate-limit metadata for sources Github, Sentry, Zendesk Support
Github: Per-hour window listed in docs.
Sentry: Docs do not explicitly list retry windows, and point to checking retry headers to determine these values. All endpoints returned a 1-minute retry window in tests.
Zendesk Support: Per-minute values listed in docs.