Recovering missing jobs/workspaces after CryoSPARC v5 database recovery – project partially restored

Data Management
10

Thank you all for your help!

I was able to delete project P6, since Delete was the only available option in our case; all of the other options were greyed out. I then successfully imported/attached the old P6 project as P73 after repairing the workspaces.json file. Specifically, I manually added entries for the missing workspaces based on the example provided here:

https://discuss.cryosparc.com/t/attach-a-project-without-workspace-json/16392/2

I also noticed that project.json contained an autodump_errors entry showing the following error:

CSModelException: Could not update mongo Job with id 699f2283e4805d7ea031da22 in collection jobs.

Update query: {'$set': {'updated_at': datetime.datetime(2026, 6, 2, 18, 15, 31, 604255, tzinfo=datetime.timezone.utc), 'dumped_at': datetime.datetime(2026, 6, 2, 18, 15, 31, 471804, tzinfo=datetime.timezone.utc), 'last_dumped_version': 'v5.0.6', 'autodump': False, 'resources_allocated.target.tpl_vars': ['ram_gb', 'job_uid', 'ram_gb_multiplier', 'job_creator', 'project_dir_abs', 'job_log_path_abs', 'job_dir_abs', 'cryosparc_username', 'worker_bin_path', 'cluster_job_id', 'run_cmd', 'num_cpu', 'project_uid', 'run_args', 'command', 'num_gpu'], 'last_updated': datetime.datetime(2026, 6, 2, 18, 15, 31, 604255, tzinfo=datetime.timezone.utc), 'last_exported': datetime.datetime(2026, 6, 2, 18, 15, 31, 471804, tzinfo=datetime.timezone.utc), 'output_results': [{'uid': 'J4267-R0', 'type': 'particle.alignments3D_multi', 'name': 'alignments3D_multi', 'group_name': 'particles_all_classes', 'title': 'Particle alignments3D multi', 'description': 'Particle alignments3D multi', 'min_fields': [('num_classes', 'u4'), ('split', 'u4'), ('shift', '({K},2)f4'), ('pose', '({K},3)f4'), ('psize_A', 'f4'), ('error', '({K},)f4'), ('error_min', '({K},)f4'), ('resid_pow', '({K},)f4'), ('slice_pow', '({K},)f4'), ('image_pow', '({K},)f4'), ('cross_cor', '({K},)f4'), ('alpha', '({K},)f4'), ('alpha_min', '({K},)f4'), ('weight', '({K},)f4'), ('pose_ess', '({K},)f4'), ('shift_ess', '({K},)f4'), ('class_posterior', '({K},)f4'), ('class', '({K},)u4'), ('class_ess', 'f4')], 'versions': [178], 'metafiles': ['J4267/J4267_00178_particles.cs'], 'num_items': [172979], 'passthrough': False}, ...}}

I inspected the J4267 job directory and confirmed that it appears to be corrupted, with several expected output files missing.

I therefore edited job_manifest.json to add the missing jobs while excluding J4267.

After these changes, the project import completed successfully and the project was assigned the new ID P73. During the import process, however, many jobs were reported as failed, particularly jobs that had been run in June and early July.

P73 still does not display all of the jobs and workspaces that were previously visible in P6 when the CryoSPARC GUI was functioning normally.

I discovered that some of the missing jobs can still be accessed directly in P73, even though they do not appear in the user interface because their associated workspaces are missing. For example, J5639 can be accessed using the following URL:

http://xxx.dbi.udel.edu:61000/#job(P73-J5639)

The job appears to work correctly. I was able to export a Result Group from P73-J5639 and then import the volume Result Group in a new project.

The remaining questions are:

  1. Is there a way to reconstruct workspaces.json so that the missing workspaces and their associated jobs can appear in the CryoSPARC user interface?

  2. Since project.json still contains the autodump_errors entry related to J4267, should anything be done to correct or remove that entry safely?