Hi again, Working on Rackham/Uppmax, running nf-core rnaseq. What do I need to do when I get

Disk quota exceeded

failure? I thought it meant that I have physically run out of storage space, but that is not the case (allocation on Crex is 1000 GB, current usage 34.7%, according to Naiss). I have, however, apparently used 760% of my 1000 core-hours for the month, so evidently I severely underestimated how much I was going to need, but it keeps going regardless?!

You might be writing to your home instead of the project storage. IIRC, storage calculation is based on the group owner of files and not the directory, so check which group your

work

and

results

directory belongs to.

You are out of free space on your project storage - currently 993 GB out of 1000 GB run

uquota

to see all numbers

That was an easier explanation...

And regarding the compute quota, you can keep launching jobs even after you've exhausted your quota, they will just have lower priority. If your jobs only request a few cores for a few hours, this might not even be a problem for you as they will fill in the gaps in the scheduler

Just to add. The stats on the SUPR web page are quite often out of sync. Should be fixed, but in general rather inaccurate for current state...

Ok, that was going to be my next question. To get more disk space, do I need to submit a continuation proposal?

You can ask support to expand your current allocation, but you need to motivate it. The allocation group will look briefly trough your data for temporary or cached data that could be deleted instead... Make sure you have removed these unnecessary files.

Hm... If I want to be able to resume, I can't remove them though, or?

can't be that one run needs 1TB - I mean potentially old runs 😉

I was running with

cleanup eager

, though I'm not sure that that actually changes anything... it didn't break anything at least

Easiest - run

ncdu

in your project folder to see where most of the space got "hidden".

Work folder is 701 GB

Wow, 701 GB in work (seems I can access the folder) and another 135 in results, the rest is insignificant. Indeed, just explain that the workflow requires unexpectedly more storage to run...

Ok, thanks! 👍

Can't comment on this

I would use

nextflow log

from where you're launching Nextflow to see past runs, you can then do

nextflow clean

to remove intermediate work-directory files for any past runs that you don't care about.

Thanks!