Error with pip package `pluggy` on cluster install

closed

#1

Hello,
I ran into an issue while installing a cluster worker (v2.5) according to the instructions. When logged into a gpu node, (through SLURM sinteractive), the installer was unable to download the pip package pluggy:

...
Retrying (Retry(total=0, connect=None, read=None, redirect=None)) after connection broken by 'NewConnectionError('<pip._vendor.requests.packages.urllib3.connection.VerifiedHTTPSConnection object at 0x7eff69c16210>: Failed to establish a new connection: [Errno 101] Network is unreachable',)': /simple/pluggy/

Could not find a version that satisfies the requirement pluggy<0.7,>=0.5 (from pytest>=2->pycuda==2018.1.1) (from versions: )
No matching distribution found for pluggy<0.7,>=0.5 (from pytest>=2->pycuda==2018.1.1)

It appears that this dependancy is not bundled with the installer, and I’m running into our cluster firewall.

Although cryosparcm does run without pluggy, the T20S tutorial fails at the motion correction step. I was able to work around this issue by running the following from a login node, from the cryosparc2_worker/ directory:

eval $(./bin/cryosparcw env)
pip install deps_bundle/python/python_packages/pip_packages/*.tar.gz --no-binary :all:

(at least the motion correction job appears to be running now, I see graphs of drift for each movie)


#2

Hi @rsrock,

I’m glad you were able to get it working. Is it possible if you can provide me with some hardware/os information so I can try to reproduce this error?


#3

Sure, here’s what we have here:

  • cryoSPARC version cryosparcm status

Current cryoSPARC version: v2.5.0

  • CUDA version cat cryosparc2_worker/config.sh, check the path to “CUDA PATH”

CUDA 9.1

  • OS (Ubuntu, CentOS, etc )

Scientific Linux release 7.4 (Nitrogen)

  • Any other relevant configurations (cluster, headless, etc )

SLURM-based cluster