Unify manifest save path #322
Merged
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
GeorgesLorre
approved these changes
Jul 27, 2023
Hakimovich99
pushed a commit
that referenced
this pull request
Oct 16, 2023
Related to #313 The local and remote runner store the manifest in different locations: * For the local runner, the manifest save path is set at compile time to a fixed path that we specify `{base_path}/{component_name}/manifest.json"` * For the remote runner (kfp), the `output_manifest_path` is set as an output artifact type. This is needed for chaining component together. The save path in this case is save internally within the VM `/tmp/outputs/output_manifest_path/data` and then mapped to minio storage after the component run (which then gets mapped to a cloud storage). The mapping saves the artifact to the specified base path followed by a fixed file structure that cannot be changed and also stored. It is also stored as zip file which contains the written text file (manifest) Example: ``` minio://soy-audio-379412_kfp-artifacts/artifacts/datacomp-filtering-pipeline-wvglp/2023/07/26/datacomp-filtering-pipeline-wvglp-3788902962/load-from-hub-output_manifest_path.tgz ``` where `/soy-audio-379412_kfp-artifacts` is the artifact bucket specified when deploying KFP. This PR unifies the manifest save path for both local and remote runner. It checks whether the `save_path` matches the expected kubeflow path and if so, save it both to the expected kfp artifact path (needed for chaining component) and the custom path that we require for caching. It's not the most optimal solution since we're writing the file twice but I don't see any other clear cut solution. @ChristiaensBert I think this might also fix some issues with the data explorer.
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.
Related to #313
The local and remote runner store the manifest in different locations:
For the local runner, the manifest save path is set at compile time to a fixed path that we specify
{base_path}/{component_name}/manifest.json"For the remote runner (kfp), the
output_manifest_pathis set as an output artifact type. This is needed for chaining component together. The save path in this case is save internally within the VM/tmp/outputs/output_manifest_path/dataand then mapped to minio storage after the component run (which then gets mapped to a cloud storage). The mapping saves the artifact to the specified base path followed by a fixed file structure that cannot be changed and also stored. It is also stored as zip file which contains the written text file (manifest)Example:
where
/soy-audio-379412_kfp-artifactsis the artifact bucket specified when deploying KFP.This PR unifies the manifest save path for both local and remote runner. It checks whether the
save_pathmatches the expected kubeflow path and if so, save it both to the expected kfp artifact path (needed for chaining component) and the custom path that we require for caching.It's not the most optimal solution since we're writing the file twice but I don't see any other clear cut solution.
@ChristiaensBert I think this might also fix some issues with the data explorer.