Closed rgreinho closed 2 years ago
Thanks for writing this up. The variety of sources that the analysis can pull its input files from (URLs, S3 cache, local files) makes it hard to keep the processing steps unified/coherent and to validate all the cases and combinations. I think this was a case where we focused on one scenario (loading the population file from a URL) and neglected the one you're trying to use.
The rm
command does do something in the case where the file is provided via PFB_POP_URL
, and I think the mv
commands are necessary in the situation where the zip filename itself doesn't match the names of the files that get extracted from it. But yeah, we can't have them crashing in the simple /data/population.zip
case. I made PR #905 as an alternative fix that I think handles both situations. Let me know if you think it will serve the purpose.
@KlaasH Thanks for your answer.
I rebuilt my image with your branch (#905) and still got an issue. I commented on the PR with the output and a potential solution.
Resolved by #905.
The image is built from
azavea/pfb-network-connectivity@0037a5c
, and the command I ran was the following:The goal is to run an analysis for the city of Metz in France, located in the Lorraine region.
In
analysis/import/import_neighborhood.sh
the block attempting to do some cleanup is erroneous and can be removed.If left, it crashes with the following error:
This is because the
population.zip
file was never moved to the tmpdir, only unzipped there, therefore there is nothing to remove.This error could be avoided by switching to
rm -f population.zip
, but then the process fails with the following error:Therfore I believe this block does not add value and can be safely removed. Once these lines where delete, the analysis completed successfully.
After that the analysis proceeds like a charm.