non-unique uuids as a problem in jobflow-remote

[JaGeo]

Hi all,

I am using:
jobflow 0.1.18
jobflow-remote 0.1.4

I have the following error message:

Traceback (most recent call last):                                                                                       │
│                File "/dss/dsshome1/00/di82tut/miniconda3/envs/autoplex_test/lib/python3.11/site-packages/jobflow_remote/jobs/run.py", │
│              line 49, in run_remote_job                                                                                               │
│                  response = job.run(store=store)                                                                                      │
│                             ^^^^^^^^^^^^^^^^^^^^                                                                                      │
│                File "/dss/dsshome1/00/di82tut/miniconda3/envs/autoplex_test/lib/python3.11/site-packages/jobflow/core/job.py", line   │
│              606, in run                                                                                                              │
│                  response.replace = prepare_replace(response.replace, self)                                                           │
│                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^                                                           │
│                File "/dss/dsshome1/00/di82tut/miniconda3/envs/autoplex_test/lib/python3.11/site-packages/jobflow/core/job.py", line   │
│              1347, in prepare_replace                                                                                                 │
│                  replace = Flow(jobs=replace)                                                                                         │
│                            ^^^^^^^^^^^^^^^^^^                                                                                         │
│                File "/dss/dsshome1/00/di82tut/miniconda3/envs/autoplex_test/lib/python3.11/site-packages/jobflow/core/flow.py", line  │
│              146, in __init__                                                                                                         │
│                  self.add_jobs(jobs)                                                                                                  │
│                File "/dss/dsshome1/00/di82tut/miniconda3/envs/autoplex_test/lib/python3.11/site-packages/jobflow/core/flow.py", line  │
│              786, in add_jobs                                                                                                         │
│                  raise ValueError(                                                                                                    │
│              ValueError: jobs array contains multiple jobs/flows with the same uuid (f7ca3f15-5957-4301-81b0-e477252e0a0f)

Any hint, idea how this could be solved? I remember running these jobs without any issue with earlier versions. Unfortunately, I don't remember the exact jobflow/jobflow-remote versions.

Or do you see this rather as a jobflow issue?

[gpetretto]

Hi @JaGeo,
just from looking at the error it would be difficult to say with certainty. However it seems that jobflow-remote does not play a big part here. After the first line of the stack trace it is already inside jobflow and it should not come back until the response is constructed.
I also think that the check that verifies that there are no jobs with duplicated uuids has been there for a long time in jobflow. I would try verifying that the same job was not mistakenly passed twice in the response. There is also the incredibly unlikely case that two equal uuids were generated alsmost simultaneously, but this would be some unbelivable coincidence...

[JaGeo]

@gpetretto thanks Guido. I am just very confused that it ran through a few weeks ago. I need to check again. Thanks in advance.

[JaGeo]

I will close this for now. It is unfortunately still not solved but I think it is likely a bug in our code. Thanks @gpetretto