You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@texera.apache.org by gi...@apache.org on 2026/08/03 23:33:54 UTC
(texera) branch gh-readonly-queue/main/pr-7242-96f47c786548c26db30efa666fdf6bb43b5406bd created (now 29e2540105)
This is an automated email from the ASF dual-hosted git repository.
github-merge-queue[bot] pushed a change to branch gh-readonly-queue/main/pr-7242-96f47c786548c26db30efa666fdf6bb43b5406bd
in repository https://gitbox.apache.org/repos/asf/texera.git
at 29e2540105 fix(workflow-operator): retry other providers on a 401 instead of aborting the HF fallback (#7242)
This branch includes the following new commits:
new 29e2540105 fix(workflow-operator): retry other providers on a 401 instead of aborting the HF fallback (#7242)
The 1 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
(texera) 01/01: fix(workflow-operator): retry other providers on a 401 instead of aborting the HF fallback (#7242)
Posted by gi...@apache.org.
This is an automated email from the ASF dual-hosted git repository.
github-merge-queue[bot] pushed a commit to branch gh-readonly-queue/main/pr-7242-96f47c786548c26db30efa666fdf6bb43b5406bd
in repository https://gitbox.apache.org/repos/asf/texera.git
commit 29e254010546b25dbac703ef59e2f1a2546772f4
Author: Prateek Ganigi <91...@users.noreply.github.com>
AuthorDate: Mon Aug 3 16:33:33 2026 -0700
fix(workflow-operator): retry other providers on a 401 instead of aborting the HF fallback (#7242)
### What changes were proposed in this PR?
In the HuggingFace operator's `_post_with_fallback`, a 401 from any
single provider immediately aborted the whole fallback chain (reported
as "Invalid HF API token"), even when another provider would have served
the model with the same token, one provider can 401 (e.g. missing
provider-specific permission) while others accept the token.
This removes the 401 short-circuit and adds 401 to the `RETRYABLE` set,
so the loop tries the remaining providers. "Invalid HF API token" is now
surfaced only when the final provider still returns 401 (i.e. every
provider rejected the token).
### Any related issues, documentation, discussions?
Closes #7194.
### How was this PR tested?
`sbt "WorkflowOperator/testOnly
org.apache.texera.amber.operator.huggingFace.*
org.apache.texera.amber.util.PythonCodeRawInvalidTextSpec"`: passes (125
tests). Added a test asserting 401 is retryable and the old
short-circuit is gone; `PythonCodeRawInvalidTextSpec` py-compiles the
generated Python. scalafmt clean.
### Was this PR authored or co-authored using generative AI tooling?
No, this PR was not authored or co-authored using generative AI tooling.
---
.../operator/huggingFace/codegen/HuggingFaceCodegenBase.scala | 4 +---
.../operator/huggingFace/HuggingFaceInferenceOpDescSpec.scala | 8 ++++++++
2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala
index 77bf488d6c..a4dfc7ca3f 100644
--- a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala
+++ b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala
@@ -195,7 +195,7 @@ object HuggingFaceCodegenBase {
| Returns (response, provider_summary). provider_summary is None on
| success or a string describing what failed.
| '''
- | RETRYABLE = (400, 404, 422, 429, 502, 503)
+ | RETRYABLE = (400, 401, 404, 422, 429, 502, 503)
| last_resp = None
| errors = []
| for prov in providers:
@@ -247,8 +247,6 @@ object HuggingFaceCodegenBase {
| continue
| if resp.status_code in (200, 201):
| return resp, None
- | if resp.status_code == 401:
- | return resp, None
| try:
| detail = resp.json().get("error", resp.text[:200])
| except Exception:
diff --git a/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDescSpec.scala b/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDescSpec.scala
index 99eb14ed6f..85e38c08e2 100644
--- a/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDescSpec.scala
+++ b/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDescSpec.scala
@@ -649,6 +649,14 @@ class HuggingFaceInferenceOpDescSpec extends AnyFlatSpec with Matchers {
outSchema.getAttributeNames.contains("hf_response") shouldBe true
}
+ it should "treat a 401 as retryable so one provider's auth failure doesn't abort the fallback" in {
+ val code = makeDesc().generatePythonCode()
+ // 401 is in the retryable set -> the loop tries the next provider instead of bailing.
+ code should include("RETRYABLE = (400, 401, 404, 422, 429, 502, 503)")
+ // The old short-circuit (return immediately on the first 401) must be gone.
+ code should not include (" if resp.status_code == 401:\n return resp, None")
+ }
+
it should "mask the API token field as a password widget in the generated schema" in {
val tokenProp = OperatorMetadataGenerator
.generateOperatorJsonSchema(classOf[HuggingFaceInferenceOpDesc])