🎉 Source S3 - memory & performance optimisations + advanced CSV options #6615
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
|
/test connector=connectors/source-s3
|
antixar
suggested changes
Oct 3, 2021
There was a problem hiding this comment.
I guess the main reason of performance issues is loading of file objects for filtering/sorting( by last_modified).
def last_modified(self) -> datetime:
"""
Using decorator set up boto3 session & s3 resource.
Note: slight nuance for grabbing this when we have no credentials.
:return: last_modified property of the blob/file
"""
bucket = self._provider.get("bucket")
try:
obj = self._boto_s3_resource.Object(bucket, self.url)
return obj.last_modified
And thus your code tries to download all files (including not relevant for incremental mode).
I propose to move this filtering/sorting logic to the bucket listing function:
def _list_bucket(self, accept_key=lambda k: True) -> Iterator[str]:
....
content = response["Contents"]
# =======
# content["LastModified"]
# =======
raise Exception(content)
except KeyError:
pass
....
This function should return relevant sorted filepath values only.
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Outdated
Show resolved
Hide resolved
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Outdated
Show resolved
Hide resolved
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Outdated
Show resolved
Hide resolved
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Outdated
Show resolved
Hide resolved
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Outdated
Show resolved
Hide resolved
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Show resolved
Hide resolved
I like the thinking but few reasons not to make that change:
I've greatly reduced mem usage (4-5x) by storing just the filepaths rather than objects, this seemed like the problem there. |
|
/test connector=connectors/source-s3
|
github-actions
bot
added
the
area/documentation
…m pyarrow ReadOptions
…3-csv-advanced-options' into george/fix-s3-oom # Conflicts: # airbyte-integrations/connectors/source-s3/setup.py # airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py # docs/integrations/sources/s3.md
|
/test connector=connectors/source-s3
|
tuliren
reviewed
Oct 16, 2021
airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Outdated
Show resolved
Hide resolved
# Conflicts: # airbyte-integrations/connectors/source-s3/source_s3/source_files_abstract/stream.py
Phlair
changed the title
Phlair
changed the title
Phlair
changed the title
|
/publish connector=connectors/source-s3
|
|
/publish connector=connectors/source-s3
|
|
/publish connector=connectors/source-s3
|
Looks like the EC2 we are using for deployment is missing |
|
/publish connector=connectors/source-s3
|
|
/publish connector=connectors/source-s3
|
schlattk
pushed a commit
to schlattk/airbyte
that referenced
this pull request
Jan 4, 2022
…ns (airbytehq#6615) * memory & performance optimisations * address comments * version bump * added advanced_options for reading csv without header, and more custom pyarrow ReadOptions * updated to use the latest airbyte-cdk * updated docs * bump source-s3 to 0.1.6 * remove unneeded lines * Use the all dep ami for python builds. * ec2-instance-id should be ec2-image-id * ec2-instance-id should be ec2-image-id Co-authored-by: Jingkun Zhuang <[email protected]> Co-authored-by: Davin Chia <[email protected]>
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.
What
closes #6606
I'm fairly confident I've been able to make some small but impactful improvements here:
How
_get_master_schema()so we can pass in the state on incremental runs and therefore only use new files to create our master schema (previously using all of them). Since the schema will be saved in state from a previous run, the only time we run the schema inference will be duringreadwhere we can pass in the state.Additionally