Add FIXMEs about handling of zstd:chunked blob annotations on blob changes #1894
Conversation
This file contains 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
mtrmac
force-pushed
the
drop-annotations
branch
from
2b4eb97
to
891630a
Compare
|
@giuseppe FYI, to make sure you have the full picture. If this handling is unreasonable, please speak up. |
|
if @giuseppe is hip with these changes, LGTM |
in the case of zstd:chunked, they apply to the binary representation. |
|
Ah, right; that sentence was describing the difficulty for the general We will probably, longer-term, end up with a list of per-concept / per-representation annotations… |
mtrmac
force-pushed
the
drop-annotations
branch
from
891630a
to
a4b6785
Compare
Signed-off-by: Miloslav Trmač <[email protected]>
…anges It's completely undefined whether the OCI blob annotations apply to the object as a concept, regardless of representation, or to the specific binary representation. So it's unclear whether we should preserve or drop them when compressing/decompressing/substituting blobs. In particular, we currently don't truly correctly handle the zstd:chunked annotations on: - decompression (should be dropped) - recompression (should be dropped) - substitution (should be replaced by data about the other blob, if any; we don't record that) Right now, we drop all annotations on decompression and recompression (which happens to work fine), and preserve annotations on substitution (which is technically invalid). Luckily, the zstd:chunked use is opportunistic, and if the annotations are invalid or not applicable, the manifest checksum fails, and we fall through to an ordinary pull; so, that is not quite a deal breaker. So, for now, just add FIXMEs recording the pain points. To fix this truly correctly, we would need: - a new metadataCleaner field in pkg/compression/internal.Algorithm - a new pkg/compression.CleanMetadata - turning public manifest.Manifest into internal/manifest.Manifest where we can add methods - adding internal/manifest.Manifest.LayerInfosWithCompression that turns MIME types into compression.Algorithm values - (using that in copy.copyLayer instead of the current hard-coded switch) - then either defining a new alternative to UpdatedImage that can handle these annotations naturally, or all the marked users that need to clean the annotations themselves calling LayerInfosWithCompression and CleanMetadata on the affected blobs. - recording the zstd annotations in the blob info cache - reading those annotations when substituting blobs based on the cache We should do all that long-term, but that's quite a lot of work to fix a metadata inconsistency which we can currently silently, with moderate cost, hide from the user. Signed-off-by: Miloslav Trmač <[email protected]>
|
@giuseppe waiting on you? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
It's completely undefined whether the OCI blob annotations apply to the object as a concept, regardless of representation, or to the specific binary representation. So it's unclear whether we should preserve or drop them when compressing/decompressing/substituting blobs.
In particular, we currently don't truly correctly handle the zstd:chunked annotations on:
Right now, we drop all annotations on decompression and recompression (which happens to work fine), and preserve annotations on substitution (which is technically invalid).
Luckily, the zstd:chunked use is opportunistic, and if the annotations are invalid or not applicable, the manifest checksum fails, and we fall through to an ordinary pull; so, that is not quite a deal breaker.
So, for now, just add FIXMEs recording the pain points.
To fix this truly correctly, we would need:
metadataCleanerfield inpkg/compression/internal.Algorithmpkg/compression.CleanMetadatamanifest.Manifestintointernal/manifest.Manifestwhere we can add methodsinternal/manifest.Manifest.LayerInfosWithCompressionthat turns MIME types into `compression.Algorigthm valuescopy.copyLayerinstead of the current hard-coded switch)UpdatedImagethat can handle these annotations naturally,or all the marked users that need to clean the annotations themselves calling
LayerInfosWithCompressionand
CleanMetadataon the affected blobs.We should do all that long-term, but that's quite a lot of work to fix a metadata inconsistency which we can currently silently, with moderate cost, hide from the user.
Also includes a fix for some typos discovered when scoping the “correct” fix.