Batching is not working

Questions
1

hi since yesterday we have some problems with http requests simply it says too many requests are made. but i didnt touch that node and it already has a batching. i saw there was a release yesterday, so could there be a problem here?

Describe the problem/error/question

What is the error message (if any)?

Please share your workflow

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

Share the output returned by the last node

Information on your n8n setup

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
2

First split two cases: external 429 versus n8n batching. In the failed HTTP Request execution, check who returns the 429 and whether there is a Retry-After or rate-limit header.

Batching only controls items inside that node execution. If several workflow executions now start at the same time, each one can still run its own batch and hit the vendor limit. Temporarily set batch size to 1 with a small wait before the HTTP Request and run one execution. If the first item still returns 429, share the redacted status/body/header and your n8n version. If only bursts fail, the fix is a queue or per-credential throttle before that HTTP node, not a release rollback.

3

but funny thing is i didnt change anything and suddenly i started to get this error. now i checked jsons, one i downloaded few weeks ago and i downloaded it again, and apparently in the new one they changed json structure to this:
“batching”: {

        "batch": {

“batchSize”: 1,

“batchInterval”: 2000

        }

previously it was:
“batching”: {

        "batch": {

“batchSize”: 1

        }

so i suspect they released something yesterday that broke it. wdyt?

4

Would help if you show your workflow.
Also a release does not mean your instance is also updated.

5

That JSON change does not prove the release broke batching by itself. It may just be the newer export writing the default interval explicitly.

The quick proof is timing: with batchSize: 1 and batchInterval: 2000, two items in the same execution should hit the HTTP node about two seconds apart. If they do, batching is working and the 429 is coming from a tighter upstream limit or multiple executions running in parallel. If they hit at the same timestamp, then you have a real batching regression. Post the n8n version and a redacted two-item execution timing; no credential or workflow secrets needed.

6

how can i see the execution timing of those http requests individually?

7

Use one tiny test execution, not the canvas preview. Open the execution, click the HTTP Request node, and compare item 0 vs item 1 in the HTTP node input/output panels.

If the UI does not show enough timing, add a temporary Edit Fields node immediately before the HTTP Request with before_http: {{ $now }} and another immediately after it with after_http: {{ $now }}. Run exactly two items and post only the n8n version plus those two redacted time pairs.

8

indeed problem was in the upstream part, apparently they did some changes and now they fixed it. so problem was not related to n8n. nevertheless thank you so much!