Architecture for sharing files across microservices

I am trying to find the best design pattern for two microservices, the first is creating a CSV file (which can get up to 1GB), the second inserts it to the DB.

I thought about 3 designs, but I can't figure out which is better:

1. upload the file using HTTP post, during the API call new message will be added to a queue, that will be consumed by a worker
2. upload the file to S3, and calling an endpoint in the other service, with file key as argument. new message will be added to the queue, worker will consume it
3. CSV service will upload a new file to S3, and write a new message directly to the queue (without posting HTTP request during the process)

please share any thoughts / ideas