Repository navigation
fix(ci): pin milvus-minio to quay.io by digest — master's Milvus job is red - #321
Merged
Merged
Conversation
…is red
MinIO stopped publishing `RELEASE.2025-09-07T16-13-09Z` to Docker Hub, so
`minio/minio:RELEASE.2025-09-07T16-13-09Z` no longer resolves and every Milvus
run dies before a single test executes:
milvus-minio Error pull access denied for minio/minio, repository does not
exist or may require 'docker login': denied: requested access to the
resource is denied
Docker Hub reports a withdrawn tag on a public repo as "pull access denied …
repository does not exist", which reads like a credentials problem and is not
one. This is on master today, so every branch cut from it inherits a red Milvus
check.
The tag is still on quay.io — where this file already gets etcd from — so the
image itself is unchanged. Pinned **by digest** rather than by tag: the failure
being fixed is a tag moving out from under us, and a tag on a second registry is
the same promise from a different party. The digest is what makes
"RELEASE.2025-09-07T16-13-09Z" mean one immutable image.
Verified locally: the image pulls, the full etcd + minio + milvus stack comes up
healthy, and `make integration-test-milvus` passes 15/15 in 84s from a clean
`docker compose down`.
(Worth noting for anyone debugging a similar failure: a *stale* Milvus container
carrying collections from an earlier session fails `test_binary_milvus_and_filter`
regardless of this change. Tear the stack down before concluding anything — the
suite is only meaningful on a fresh one, which is what the make target does.)
Split out of #319, where it was riding along with unrelated dataset work.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Master's
Integration tests (Milvus)job is currently red, and has been since MinIO withdrew this release tag from Docker Hub. Every branch cut from master inherits it — e.g. #320.Docker Hub reports a withdrawn tag on a public repo as "pull access denied … repository does not exist", which reads like a credentials problem and isn't one. The job dies at
docker compose upbefore a single test runs.The fix
The image is still published on quay.io — which this compose file already uses for etcd — so the image itself is unchanged.
Pinned by digest, not by tag:
The failure being fixed is a tag moving out from under us. A tag on a second registry is the same promise from a different party, so re-pinning to another tag would just relocate the risk. The digest is what makes
RELEASE.2025-09-07T16-13-09Zmean one immutable image.Verification
etcd + minio + milvusstack comes up healthymake integration-test-milvus→ 15 passed; 0 failed, 84.23s, from a cleandocker compose downOne debugging note worth having in the record
A stale Milvus container — one carrying collections from an earlier session — fails
test_binary_milvus_and_filterregardless of this change. I hit exactly that and nearly mis-attributed it to the fix. Tear the stack down before concluding anything; the suite is only meaningful on a fresh one, which is what the make target does.Split out of #319, where this was riding along with unrelated dataset work. It has no dependency on that PR and unblocks master on its own.
🤖 Generated with Claude Code