It seems that the server disk space is not large enough to hold docker instances.
Could anybody solve this problem? For example, simply enlarge the server disk space, or optimize the docker instance scheduling algorithm.
Error logs are as follows.
Unable to find image 'scaelles/codalab:anaconda3-2018.12' locally
anaconda3-2018.12: Pulling from scaelles/codalab
ff4229790957: Pulling fs layer
d2a3b85993ea: Pulling fs layer
388da27c8816: Pulling fs layer
271b153e4221: Pulling fs layer
381ecf3648aa: Pulling fs layer
d5c134fc2997: Pulling fs layer
271b153e4221: Waiting
381ecf3648aa: Waiting
d5c134fc2997: Waiting
d2a3b85993ea: Verifying Checksum
d2a3b85993ea: Download complete
ff4229790957: Verifying Checksum
ff4229790957: Download complete
271b153e4221: Verifying Checksum
271b153e4221: Download complete
381ecf3648aa: Download complete
ff4229790957: Pull complete
d5c134fc2997: Verifying Checksum
d5c134fc2997: Download complete
d2a3b85993ea: Pull complete
388da27c8816: Verifying Checksum
388da27c8816: Download complete
docker: failed to register layer: Error processing tar file(exit status 1): write /.pyenv/versions/anaconda3-2018.12/pkgs/scipy-1.1.0-py37h7c811a0_2.tar.bz2: no space left on device.
See 'docker run --help'.
Hello,
Thanks for letting us know, we have contacted the Codalab admins as the pool of workers is managed by them.
For now, the best solution is to submit again as it usually succeeds at the second or third attempt.
Best,
Sergi Caelles