Hello,
Is there a way of hosting an sheduling targets on Connect? How is sandboxing affecting the files written to disk?
Can you provide a bit more detail on what you are trying to accomplish?
Additionally are you doing this in a Connect server with off-host execution enabled or Connect native? Sandboxing is a bit different depending on how the server is configured.
I execute nativ on connect. After reading the docs, I managed to host and launch a targets via a quarto document, although data need to be stored on a S3 object store because connect is sandboxing the deployed content in a way that any files written to disk during execution are discarded. Regards
It sounds like you are launching a targets pipeline from a Quarto doc.
If that pipeline writes data to disk, the job running the Quarto document could be getting terminated which means the written data gets removed. If you are trying to access the data from another piece of content it can't due to sandboxing.
targets by default stores intermediate computation state on disk. However it can be configured to write to cloud storage. See the targets documentation on Cloud Storage.
Pushing up to S3 and pulling the data down sounds like a good solution, but I can recommend some other resources.
You may be able to use pins. We have a Cookbook document called "Workflow Automation in R using targets" that may be helpful, that publishes the results to a pin.