[Terra_2_NCBI] Disrupted NCBI Connections Often Cause Failures

[michellescribner]

🐛

📝 Describe the Issue

Terra_2_NCBI is one of the most frequent workflows where troubleshooting support is requested. A common cause is unexpected disrupted connections to the NCBI FTP server as indicated in the stdout file below.

https://storage.cloud.google.com/fc-4273801f-4c61-4f2a-861a-5d62ffaad5e6/submissions/2d734794-a7b6-435d-9594-ebfd3b4b647f/Terra_2_NCBI/2f9488c6-d8d5-4835-9a7c-42c16beadcd6/call-biosample_submit_tsv_ftp_upload/stdout?authuser=0

I suspect that a contributing factor to this issue is that during BioSample submission, the tasks queues the samples for submission, then waits for a response back from NCBI with the BioSample accessions. For example, the task waited for 2 minutes in the example submission above.

This is not a straightforward fix, but could we configure the parameters of the task such that connection disruptions will be less frequent? Alternatively, could we decouple the BioSample submission from gathering the accessions?

🔁 How to Reproduce

💻 Version Information

[sage-wright]

Unfortunately, this isn't a fix we can do without altering the underlying Asymmetrix script that is being used. Add this to the list of reasons why we need to rework this workflow from scratch imo