Skip to content

Update resource requirements for LoTSS subtraction.#87

Open
tikk3r wants to merge 4 commits intomasterfrom
slurm-resources-subtraction
Open

Update resource requirements for LoTSS subtraction.#87
tikk3r wants to merge 4 commits intomasterfrom
slurm-resources-subtraction

Conversation

@tikk3r
Copy link
Member

@tikk3r tikk3r commented Feb 11, 2026

This PR adds resource requirements based on running the subtraction on a Lockman pointing on cosma.

DIS2 h5parm merging

87.61user 63.33system 2:34.87elapsed 97%CPU (0avgtext+0avgdata 86228096maxresident)k                     │
8226448inputs+28627944outputs (1725major+1981817minor)pagefaults 0swaps

Subtracting one chunk
Looking at the output of time it seems the prediction only uses ~45 GB

3589.76user 403.85system 12:43.18elapsed 523%CPU (0avgtext+0avgdata 45454272maxresident)k                │  - class: ResourceRequirement
2143544inputs+100504136outputs (2932major+52721951minor)pagefaults 0swaps  

the the OoM killer kept kicking in and DDFacet itself reported ~125GB so perhaps there's a small spike somewhere or something like that. That is where the 128 GB ramMin comes from.

@tikk3r tikk3r linked an issue Feb 11, 2026 that may be closed by this pull request
@tikk3r tikk3r requested a review from jurjen93 February 11, 2026 11:28
- entry: $(inputs.ms)
writable: true
- entry: $(inputs.solsdir)
writable: true
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why does this needs to be writable? Given that we ran this successfully without making it writable?

Copy link
Member Author

@tikk3r tikk3r Feb 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I and Gabriella couldn't run it successfully without that folder being writable recently for some reason. I'm not sure why.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It comes at a cost of an extra (maybe unnecessary) copy. Maybe worthwhile to figure out why this is suddenly needed.

Copy link
Member Author

@tikk3r tikk3r Feb 13, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

New run is running that has the MSes in SOLSDIER renamed to the pipeline outputs instead of symlinked to the ddf outputs. Let's see if that works better.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Renaming the ddf-pipeline's output instead of adding a symlink to them has worked, so for some reason it doesn't like the symlink. We can just rename them before running the workflow, so I have removed the writable flag for SOLSDIR.

@tikk3r tikk3r force-pushed the slurm-resources-subtraction branch from f49ec4b to cd2662b Compare February 12, 2026 13:49
@tikk3r tikk3r force-pushed the slurm-resources-subtraction branch from cd2662b to 33ac785 Compare February 13, 2026 13:56
@tikk3r tikk3r requested a review from jurjen93 February 13, 2026 14:19
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Update memory requirements for delay-calibration workflow

3 participants