Error: Resource temporarily unavailable about nanocaller HOT 4 CLOSED

Hi Neng,

I am assuming that you are using NanoCaller_WGS.py to run NanoCaller because that's where I have seen this problem before. This is happening because NanoCaller is running 48 tensorflow sessions, one in each thread, which can exceed the number of processes allowed.

Here are a few suggestions that can help:

1. Run ulimit -u unlimited command in terminal as suggested here: https://stackoverflow.com/questions/52026652/openblas-blas-thread-init-pthread-create-resource-temporarily-unavailable.
2. Use fewer threads than 48, perhaps 16 or 24.
3. Use docker container to run NanoCaller. I have noticed that the problem you mentioned goes away when I use docker to run NanoCaller instead.
4. Use NanoCaller.py command instead for each contig individually. If you use any number of CPUs with this command, it creates only one tensorflow session and uses other threads for feature generation only.

I am looking at reorganizing/reimplementing parallelism in NanoCaller_WGS.py in a way that it also creates only one tensorflow session but that might take some time.

Let me know if this helps,
Umair

from nanocaller.

Hi Umair:
I am trying to run NanoCaller on HPC using singularity. After I pulled the image and started to run as the steps you provided, it told me can not find the file.

from nanocaller.

Hi,

I think the problem is that NanoCaller is installed in /app inside to Docker container and by default the working directory is /app. But singularity ignore this.

Try the following:

singularity run nanocaller_${VERSION}.sif python /app/NanoCaller.py --help

or

singularity exec -e --pwd /app nanocaller_${VERSION}.sif python NanoCaller.py --help

Let me know if this works out. Also, please use the version 1.0.1 for NanoCaller because it has some speed up optimizations, I have just uploaded a docker image for it.

from nanocaller.

To umahsn:
It works when I tested the two command you provided. Thanks a lot.

Best
Neng

from nanocaller.