Mirror upstream elastic/elasticsearch#133344 for AI review (snapshot of HEAD tree) #210
Conversation
BASE=a31485f6e8f14869de0605e9f6b303b353b772a0 HEAD=be3c1d34ffb0c20f2517b1b81283f7bd311024ac Branch=main
|
@codex review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| String previousPath = null; | ||
| for (int i = 0; i < numberOfValues; i++) { | ||
| Dimension dim = dimensions.get(i); | ||
| String path = dim.path(); | ||
| if (path.equals(previousPath)) { |
There was a problem hiding this comment.
The value-similarity loop iterates only numberOfValues = min(MAX_TSID_VALUE_FIELDS, dimensions.size()) entries, but array elements are skipped with continue. If a time series starts with many values for the same path (e.g., an array with >16 items) the loop exhausts its numberOfValues budget before reaching later unique dimensions, so those fields never contribute a value byte to the TSID. Different series that vary only in those later fields will therefore hash to the same value-similarity prefix and cluster together, defeating the stated goal of clustering by up to 16 field values. Consider iterating until 16 distinct paths are processed rather than blindly stopping after the first 16 elements.
Useful? React with 👍 / 👎.
This is a part of elastic#132566