Skip to content

Improve TMPDIR use #2

@moskalenko

Description

@moskalenko

Hi, Carl,

On HPC clusters /tmp is often unusable for analyses, so, unfortunately, your package is causing compute node downtime because of how it calls external tools and uses /tmp.

Please modify any code that uses /tmp directly to use the $TMPDIR path for all tools by default. R call 'tempdir()' honors $TMPDIR as far as I can see.

If you can, please also provide the users of the package with an option to specify the temporary directory path when calling functions from the package.

Since it looks like you're calling some tools directly via jars the "java -Djava.io.tmpdir=SOMEPATH" argument pointing to the value of $TMPDIR should also work reasonably well for tools like gatk.

Some of the hardcoded paths found in the code:

PhyloProcessR $ $ grep -r '/tmp'
R/assembly.assembleSpades.R: system(paste0("rm -r ", save.assem, "/tmp"))
R/assembly.assembleRNASpades.R: system(paste0("rm -r ", save.assem, "/tmp"))
R/preprocess.normalizeReads.R: system(paste0("rm -r ", out.path, "/tmp"))
R/variants.mapReferenceSample.R: tmp.dir <- paste0(lane.dir, "/tmp")
R/variants.mapReferenceConsensus.R: tmp.dir <- paste0(lane.dir, "/tmp")

Thank you!

Alex

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions