Does multiqc somehow create less plots in the `mul...
# multiqc
j
Does multiqc somehow create less plots in the
multiqc_plots
folder dependent on the nextflow version? πŸ€” I am a bit confused about this one (might also be a pipeline issue) https://github.com/nf-core/mhcquant/actions/runs/18346526475?pr=401
p
I don't think that MultiQC knows or cares about Nextflow
I mean, multiqc ❀️ nextflow of course, don't want to hurt any software's feelings
πŸ˜† 2
nextflow deal with it 1
But MultiQC doesn't alter its behaviour in any way according to whether it is or isn't run within a Nextflow pipeline
j
Thanks phil πŸ˜„ what I thought as well. Guess I’ll dig a bit more into it… this is fine fire
this is fine fire 1
p
Actually looking at the CI logs now. Looks like new plots,
mass_error.{pdf,png,svg}
and
scores_xcorr.{pdf,png,svg}
right?
They are showing up as new in the CI logs for both versions of Nextflow though, so nf-test is throwing an error with Nextflow latest:
Copy code
FAILED (550.442s)

  Assertion failed: 
  
  1 of 2 assertions failed
  
  Nextflow stdout:
  
  N E X T F L O W  ~  version 25.04.8
it's just that those CI jobs are showing as passed, when they should be failed. Which is... odd
Looks like it's the step that checks if nf-test failed, which is itself failing, which is causing the CI test to pass πŸ˜΅β€πŸ’«
Copy code
Error: Test with latest-everything failed
Warning: Test with latest-everything failed but will not cause workflow failure. Please check if the error is expected or if it needs fixing.
aaahhhh 1
unclear why it failed, it looks fine πŸ€”
Copy code
if [[ "failure" == "failure" ]]; then
    echo "::error::Test with latest-everything failed"
    # Add to workflow summary
    echo "## ❌ Test failed: docker | latest-everything | Shard 1/5" >> $GITHUB_STEP_SUMMARY
    if [[ "latest-everything" == "latest-everything" ]]; then
      echo "::warning::Test with latest-everything failed but will not cause workflow failure. Please check if the error is expected or if it needs fixing."
    fi
    if [[ "latest-everything" != "latest-everything" ]]; then
      exit 1
    fi
  fi
ohhh wait it's the
latest-everything
thing. This was a change that we made not long ago, where test failures on Nextflow
latest-everything
don't cause the nf-core CI to fail
πŸ’₯ 1
So that new edge releases can't take down all nf-core CI in one go without warning
Ok so this makes sense to me now: β€’ Your pipeline added some new MultiQC plots β€’ All nf-test CI jobs failed β€’ The
latest-everything
ones show as passed because that's nf-core policy for CI testing So if you update the nf-test snapshots to include the new plots, they should all go green and you should be fine to continue.
We should probably supplement this log message with a link to some docs:
Copy code
Warning: Test with latest-everything failed but will not cause workflow failure. Please check if the error is expected or if it needs fixing.
this 1
j
Thanks for debugging! That makes sense then πŸ™ Updating the snapshots should fix this. ACTUALLY if I have your attention now πŸ˜„ I would want to update snapshots with nf-core bot (see this PR). We could test this on pipeline level right away
p
oh nice! Yeah seems like a good test.
Can you start a new thread in #C069B2Z5P1R about this first please?
πŸ‘ 1
I went on a huge wild-goose-chase trying to build this the other day (trying to be too clever and using snapshot files from past runs, was a failure). And Julia has also built other workflows specifically for GPU etc.
It'd be good to get eyes on it quickly before we merge + test
πŸ™Œ 1
If it works I'd love to add it into the template. Might even get it in before this tools release if we're speedy.
If you post I can ping fokls
j
Posted βœ…
p
Thanks!