Closed nicholasbalasus closed 4 months ago
Yes we're on PBSPro and I also notice I have to go through different files for error handling sometimes. Also it might be the case that logs are named something other than "imi_output.log" if running multiple inversions, so the name being hardcoded can cause some confusion.
This issue has been automatically marked as stale because it has not had recent activity. If there are no updates within 7 days it will be closed. You can add the "never stale" tag to prevent the issue from closing this issue.
@nicholasbalasus @sabourbaray Is this still an issue or have we resolved it? I have noticed sometimes you need to look in the stderr output (e.g. the slurm output file on our system) to find error messages.
Hi Melissa - I think this is still an issue. I'll take a shot at fixing it.
Closed by #199
Name and Institution (Required)
Name: Nick Balasus Institution: Harvard University
Description of your issue or question
This is low priority. Working from the
dev
branch, I find that errors in the IMI preview (specifically fromimi_preview.py
) are not output toimi_output.log
as intended (code that should address this below).https://github.com/geoschem/integrated_methane_inversion/blob/9d6a45a1eaf355c00e6bd63686957317905fcca8/src/inversion_scripts/imi_preview.py#L99-L101
For example, I add a divide by zero error here:
https://github.com/geoschem/integrated_methane_inversion/blob/9d6a45a1eaf355c00e6bd63686957317905fcca8/src/inversion_scripts/imi_preview.py#L60
The preview will fail. However, the divide by zero error is nowhere to be found (it is not in the slurm-*.out files in the PreviewDir). The only indication in
imi_output.log
is that something has a gone wrong (snippet below), but we can't see what.This is a general problem for all of our scripts called with sbatch, as they will typically print the error to a slurm file. The reason imi_preview is not sending them even to the slurm file is (I think) related to the above error handling code being inside a function (and not the function where I put the error-causing code). Happy to put this in a future PR if anyone has a recommended solution.