instaGRAAL on usegalaxy.eu finishes as “ok” but all outputs are 0 bytes

usegalaxy.eu support
, , ,
1

Hi!

I am reporting a likely problem with the instaGRAAL tool on usegalaxy.eu.

I ran instaGRAAL (0.1.6+galaxy0) with these inputs:

  • abs_fragments_contacts_weighted.txt

  • fragments_list.txt

  • info_contigs.txt

  • reference genome FASTA

The job ran for about 33 hours and Galaxy marked it as ok, but all 3 output datasets are 0 bytes:

  • Genome sequence

  • Fragment information

  • Curated sequences

In the Job Information page I see:

  • Tool Exit Code = 1

  • Job State = ok

  • Job message:

    • Warning: instaGRAAL is raising an expected error at the end.

    • error_level: 2

    • type: regex

    • stream: stderr

    • match: instagraal_class

This same behaviour happened in two independent histories with different datasets, so it seems systematic and not input-specific.

The output dataset peeks do not contain real FASTA/tabular results, and the datasets remain at 0 bytes even after refresh.

Could you please check whether the Galaxy wrapper is masking a real failure or failing to collect the expected output files?

Thank you.

2

Welcome @alma.wis

Hopefully we can help!

I’m running a simple test on the server right now.

Meanwhile, have you checked the full job logs? This is where to look → FAQ: Troubleshooting errors.

I would also suggest that you send in a bug report from each error job (one dataset from each). You can include a link to this topic in the comments. The administrators of server screen these messages.

You can copy paste the logs back here if you would like to!

More soon as I run my test! :slight_smile:

3

@alma.wis we are aware of this issue with the tool and a few of us have tried to fix it.

Not 100% sure what the latest state is, but obviously things are still not working as they should.

I’m sorry, but I don’t have a better anwer for you at the moment than that this tool is currently unusable and that you may check back here in a few weeks to see if things have changed. I’ll post here when that’s the case.

4

@jennaj marked this as solved to have it show up on the tool interface.

5

Hello,

I am trying to run instaGRAAL on Galaxy, but the job fails during the PyCUDA/NVCC compilation step.

The error is:

nvcc fatal : Value ‘sm_89’ is not defined for option ‘gpu-architecture’

The input files seem to be processed correctly because the pyramid is built before the crash. The failure happens only when PyCUDA tries to compile CUDA kernels.

This suggests that the GPU has compute capability sm_89, but the installed CUDA/NVCC version does not support it. Could you please update CUDA/NVCC, or route the instaGRAAL job to a GPU architecture supported by the current CUDA version?

Thank you.

6

Hi @alma.wis

Thanks for sharing the extra details!

The EU team will see your message. They are aware of the issue and will try to resolve it soon! We only marked this as “resolved” to help others that run into the same issue to find the current status.

I’ll mark this for an update in a few weeks, and please feel free to write back for an update if you don’t hear from us after the first week in June.