I have recently moved from NXF V23 to v25 (server ...
# help
n
I have recently moved from NXF V23 to v25 (server based) and am seeing a lot more files created and disk space used when running the pipeline. I can't seem to get a definitive answer as to whether this is really a result of the upgrade or something else in my pipeline. Anyone have any ideas?
p
Can you elaborate on what the files are?
But no, nothing springs to mind from the Nextflow side here.
n
Not sure really, the work directory for a run of 1 million id's contained 50,992 files (40G), but we had previously seen <15,000 for this number of id's. Seqera.A.I seemed to suggest there were more metadata files with v25 but I'm not sure how true that is.
p
Can you compare two task dirs and see what's different?
Have you enabled Nextflow Lineage? That could create additional JSON files, but I doubt so many and they're not big
n
Unfortunately the work directory with the 15K files is long deleted (we have a tight inode limit). I haven't knowingly enabled Lineage. I just wondered, as we had done quite a lot of work to reduce intermediate file numbers and the inode count and disk space had been very manageable. Then, all of a sudden we saw a massive increase in inodes and, for one run, 230G of disk space use. We then switched to using
scratch = true
to combat this and it certainly decreased. All this seemed to correlate with a move from
v23
to
v25
but, of course, there may be other reasons.
p
Fair enough. I would probably try a small diagnostic run with both versions to see if you can replicate it and narrow it down.