contentutil: avoid defaulting to ReadAt for fetch #6366
Merged
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Outdated suggestions cannot be applied.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.
Containerd v2.2 updated the ReadCloser returned from Fetcher to implement ReadAt, but it works by making a separate HTTP request for each read. This means lots of requests depending on the buffer size used by copy. In containerd content pkg buffer is 1MB.
Containerd change containerd/containerd@4bf1705
@cpuguy83 @dmcgowan I think this is unexpected that the objects return common interfaces but their performance may differ multiple times (100x+ more requests). This seems to fix the main pull case for buildkit, but there may be more accidental cases for hitting the new
ReadAtmethod assuming it works with similar performance asRead/Seek. Lots of things likeSectionReaderuseReadAtby default and containerdcontent.ProviderisReadAtbased.