My requests keep being rate limited, where's the limit and why do I keep hitting it? #3944
Replies: 3 comments 4 replies
-
GitHub has secondary rate limits that you may be triggering. Does the same thing occur if you only check one package at a time? As for the logging, those messages should be enabled by default. Can you paste your output that you are seeing? |
Beta Was this translation helpful? Give feedback.
-
No, it does not occur on a single run - but I would like (=need) to check multiple packages or know about the feasibility to do so.
Is the output of a single run, which already ran for 30s. I get it - hitting the secondary rate limits (which I read of) would lead to the observed behavior, but I'm blind and I don't like that... by enabled by default, what exactly do you mean? I ran (and I know that show-results doesn't help me, verbosity might help me, but didn't yet) Edit: one error passed through earlier on:
but right now I was looking at a stalled execution again, no logs for 8 minutes. My mix up of parameter style should be ugly, but not the cause (foo.json has been filled properly). |
Beta Was this translation helpful? Give feedback.
-
I had another run at the topic. I discovered using another token type ("classic") seems to perform better than the PAT I configured earlier. Additionally, there seem to be notable differences between "latest", tagged docker container and my homebrew install (which should be 4.13.1), but I don't want to burn my token/account before having solved the logging issue. Funnily enough, as I wrongly called docker from within python ("-e var=var" as single arg), scorecard complained correctly and logged that including a rate limiting message). |
Beta Was this translation helpful? Give feedback.
-
Hi,
I can't figure out how to debug this (I'm not familiar with go and logr), and I keep getting growing delays inbetween my requests. I wrote a small python wrapper to check multiple packages at once and it looks to me like github was rate limiting me, ending up with scorecard execution durations from 3, 5, 10, 30, 90, ... seconds, until I cancel requests by timeout.
I believe this doesn't match 5k messages per hour - although of course I don't know how (often) scorecard queries github for the queried information. I skipped the "Vulnerabilities" check already, as it queries another SaaS service.
I found this issue already, but the actual question wasn't answered... though it would help me to figure out how to print this log 😄
#2945
Beta Was this translation helpful? Give feedback.
All reactions