NERSC adapter: POST /compute/status (job listing) timeouts

[osmiumzero]

Summary

POST to /compute/status for listing all jobs times out. Individual job status queries (GET /compute/status/{resource_id}/{job_id}) work fine.

Impact

Minor — individual job status checks work, so this only affects bulk job listing. Workaround is to track job IDs client-side and query them individually.

Environment

• NERSC Perlmutter
• API: https://api.iri.nersc.gov/api/v1
• Date: 2026-03-03

[gabor-lbl]

Yes, unfortunately, this will be the case until we move off of the current globus token-based implementation. In order to make globus tokens work, I had to disable the use of our fast-responding jobdb integration. Once we have a real AmSC token implementation, we can add the jobdb back (which can report on thousands of jobs in milliseconds.) Thanks for logging this!