Potential bug in tmp directory cleanup

[edwhisnant]

Hey Jon,

I have recently noticed a weird bug after I run the f2 pipeline where my input genome assemblies are getting deleted after the f2 run has finished. I have scoured my own wrapper code to make sure I am not inadvertently removing my files. So far, I have not identified a source of the deletion. However, I noticed that in my input directory there were several temporary f2 clean directories that were in still in progress. I wonder if during the temp file cleanup, the input files are getting removed?

I have now started to send my temp files into a dedicated temp directory via --tmpdir for the clean, predict, and annotate steps to ensure that it is not the temp dir cleanup that is causing this. I'll let you know if this solves my issue.

I just wanted to put this on the repo in case someone else has run into this issue (or might run into this issue in the future).

[edwhisnant]

Okay, so I verified that this was not a bug of my own writing (I wasn't using rm, mv, or some other method of overwriting/deleting) and verified that my scripts were only "reading" (not modifying) the input fasta files. This is extremely weird, because I have used these scripts in the past and never ran into this issue. So, anyway, one of the following seems to have worked:

1. I had to run an early cd in my script to move into a random empty /tmp directory to ensure it was not because I was executing f2 in the input directory
2. Changed the directory and input file permissions to read only (I should see a Permission denied or some other verification before a mv or rm is executed).
3. Made the tmp file output explicit in funannotate2 using --tmpdir

[nextgenusfs]

Was the whole output folder being deleted or just the files? I can't recall anywhere where we are doing cleanup steps that would cause this. But if reproducible it would help to let me know the commands/context so I can fix.

[edwhisnant]

No, what is really odd is that it was not the output. it was my input directory with the genomes that I am feeding into funannotate2 that was getting deleted. I would start a batch of genomes and when I returned a few hours later the entire directory would be removed, with the only files being left being clean-######. This is why I suspected it may have been a potential bug in the temp file cleanup.