Hi all. Just a quick question. I am running the nf...
# nostupidquestions
b
Hi all. Just a quick question. I am running the nf-core/mag pipeline. Sometimes a job task will fail and I'll go to the working directory for that task to investigate. However, the input files won't be in that directory. I assume they're in another directory within the work/ directory, since the input files are created by a different upstream task. Is there an easy way to trace back and find them? Thanks! 🙂
j
Depending how at what scale you need to track this, when I'm in a particular work directory I just look at the output of
ls -l
which should tell you the source of the symlinked input file
Alternatively there is the native Nextflow
lineage
functionality which might also give you such information
b
Hey James, thanks for the reply. I thought a bit more and I think the problem is something to do with the process.scratch = true option in my config file. I enabled this to take advantage of the node storage (Im on an HPC) rather than writing to the disk. Thus, the symlinks are getting written to /scratch on the node rather than in the work dir first.
j
Ah got you
Yeah in that case you woun't have symlinks as I think the files get copied in
I think the data lineage thing may help with that
Also possibly if you run with the
debug
profile additionally, but this will produce a big scary log 😅
One other option:
You can use the table in the
pipeline_info/execution_report<...>
Search for the sample that failed
And find the upstream step (if you know which one it is). Presumably it woud be linener in the table
b
Wow, thanks for the thorough help! 'll look into your suggestions 🙂
j
For one-off stuff the pipeline table is probably easiest 🙂
👍 1