fix(OpenAI Node, Basic LLM Chain Node, Tool Agent Node): Better OpenAI API rate limit errors

[burivuhster]

Summary

In this PR some custom handling for the OpenAI rate limit errors is introduced.
According to OpenAI's docs, there might be two sets of errors ending up with 429 HTTP response: rate limit errors and insufficient quota errors (https://platform.openai.com/docs/guides/error-codes/api-errors)
We can look at the code in the response to see what kind of error it was.

Also, we override a default message for those errors with a more descriptive and openai-specific ones.
This should also help with the issue, when a long poorly formatted text is displayed in some cases of rate limit errors.

Related Linear tickets, Github issues, and Community forum posts

https://linear.app/n8n/issue/AI-313/openai-is-returning-429-errors-quite-often
https://linear.app/n8n/issue/AI-342/bad-error-in-openai-node

Review / Merge checklist

• PR title and summary are descriptive. (conventions)
• Docs updated or follow-up ticket created.
• Tests included.
• PR Labeled with release/backport (if the PR is an urgent fix that needs to be backported)

[jeanpaul]

Left a few comments to clarify whether this is actually only triggered by OpenAI models or not.

[jeanpaul]

How do you know, at this point, that it's an actual OpenAI error, and not some other API error that happens to have an error-code? Please check if this check is correct, and/or if the name of the variable could signal something better :)

[burivuhster]

Thanks, I've added a check that the underlying error comes from OpenAI

[jeanpaul]

Some models might be blocked by the policies of a project too, which results in a weird error "Forbidden":

  "errorMessage": "Forbidden - perhaps check your credentials?",
  "errorDescription": "Project `proj_r4lQnxwlsEgGpJiweSPiNTfO` does not have access to model `gpt-3.5-turbo`",
  "errorDetails": {
    "rawErrorMessage": [
      "403 - {\"error\":{\"message\":\"Project `proj_r4lQnxwlsEgGpJiweSPiNTfO` does not have access to model `gpt-3.5-turbo`\",\"type\":\"invalid_request_error\",\"param\":null,\"code\":\"model_not_found\"}}"
    ],
    "httpCode": "403"
  },
  "n8nDetails": {
    "nodeName": "OpenAI",
    "nodeType": "@n8n/n8n-nodes-langchain.openAi",
    "nodeVersion": 1.5,
    "resource": "text",
    "operation": "message",
    "itemIndex": 0,
    "time": "9/17/2024, 4:51:36 PM",
    "n8nVersion": "1.59.0 (Self Hosted)",
    "binaryDataMode": "default",
    "stackTrace": [
      "NodeApiError: Forbidden - perhaps check your credentials?",
      "    at Object.requestWithAuthentication (n8n/packages/core/src/NodeExecuteFunctions.ts:1924:10)",
...
    ]
  }
}```

[jeanpaul]

Would be cool if you could add the "forbidden" error to the error messages too!

[github-actions]

✅ All Cypress E2E specs passed

[cypress]

n8n    Run #6935

Run Properties:   Passed #6935  •  780cb8dbce: 🌳 🖥️ browsers:node18.12.0-chrome107 🤖 burivuhster 🗃️ e2e/*

Project	n8n
Branch Review	ai-313-openai-is-returning-429-errors-quite-often
Run status	Passed #6935
Run duration	04m 43s
Commit	780cb8dbce: 🌳 🖥️ browsers:node18.12.0-chrome107 🤖 burivuhster 🗃️ e2e/*
Committer	Eugene Molodkin
View all properties for this run ↗︎

---

Test results
Failures	0
Flaky	3
Pending	0
Skipped	0
Passing	430
View all changes introduced in this branch ↗︎

[janober]

Got released with [email protected]