Q: Is the main branch latest objects directly accessible directly from underlying object store (s3) without using lakefs? what approaches have you guys taken wrt. to something like this. eg. I would like to ingset data to snowflake using snowpipe or make the objects available to any other service that does not have the opportunity to communicate with lakefs

Hi @Leonard Aukea, The straight forward way to make the object available for users that can access lakeFS is to export the data https://docs.lakefs.io/reference/export.html. The disadvantage of doing that is that it requires copying data. We currently have an issue using the export functionality from Databricks but other methods would work. If you are not willing to copy data, you can use

lakectl fs stat

to get the physical address of each object you are willing to expose.

Yeah. preferably I would not want to copy the data. Just thought it could have made sense to expose the latest version of main branch so that other systems could easily make use of it. So how would I go about running

lakectl fs stat

to get the physical adress of each object in a repository for a particular branch and commit?

The flow of the script you would want to implement is: • list the objects under the branch you are trying to expose (e.g.

lakectl fs ls --recursive <lakefs://example-repo/master/>

) • for each object get its stats (e.g.

lakectl fs stat <lakefs://example-repo/master/file>

) • parse the stat to get the physical address. You can use one out of two methods to implement this flow: • Assuming that you have lakectl installed, or you are planning to install it Write a bash script that uses lakectl operations • You can use our Python client that supports the same operations. Please let me know if you have additional questions