Issue 87: ALCF Recon Flow

[davramov]

Simple fixes to the ALCF recon flow as outlined in this issue: #87

Update: also adding a timeout to globus compute jobs to address this issue: #45

[kkamdin]

Hey David, I don't see any big issues so this will probably be fine as-is. I flagged two things in comments that you may want to check out in bl823/alcf.py before merging:

• mixed logging modules
• if/else and ValueError raise behavior in the flow steps

[kkamdin]

does Step 2B (the multi-res flow below) depend on data832_tiff_transfer_success being true? If yes, you could wrap 2B in an if/else statement so it doesn't try to run and crash.

Or do a similar value error raise if that's the right behavior like you do here (if I have this right -- if alcf_multi_res_success is not true you don't try to run the zarr transfer "Transfer B" even further below):

            if not alcf_multi_res_success:
                logger.error("Tiff to Zarr Failed.")
                raise ValueError("Tiff to Zarr at ALCF Failed")
            else:
                logger.info("Tiff to Zarr Successful.")

[davramov]

The multiresolution step does not rely on the TIFF transfer being successful, just that reconstruction was successful. I switched the transfer order so that users have the option to view the TIFF stack at the beamline before waiting for the Zarr conversion to complete (an extra 10+ minute overhead).