It does match the image.
tensorflow==2.4.0+nv should be changed to tensorflow==2.4.0
But for some reason although all the gpus are free and no other agent is on the machine, only one task is executed at the time instead of 4. Why is that?
Can you make sure the agent did receive the GPUs 0 through 3?
ImmensePenguin78 this is probably for a different python version ...
SuccessfulKoala55
Well, I've removed the requirement altogether and now it won't fail on this anymore (TF is provided anyway AFAIK via the image) but now I get the following:
Any ideas?
*Needless to say, when running locally this works with no problem. Also the http://nvcr.io/nvidia/tensorflow:21.02-tf2-py3 image is able to run TRT
Hi ImmensePenguin78 , you mean you want to run other copies of the same task using different docker images?
Well the requirements were automatically filled, not by me
SuccessfulKoala55 On another note, I'm also getting
ERROR: Could not find a version that satisfies the requirement pandas==1.3.4 (from versions: 0.1, 0.2, 0.3.0, 0.4.0, 0.4.1, 0.4.2, 0.4.3, 0.5.0, 0.6.0, 0.6.1, 0.7.0, 0.7.1, 0.7.2, 0.7.3, 0.8.0, 0.8.1, 0.9.0, 0.9.1, 0.10.0, 0.10.1, 0.11.0, 0.12.0, 0.13.0, 0.13.1, 0.14.0, 0.14.1, 0.15.0, 0.15.1, 0.15.2, 0.16.0, 0.16.1, 0.16.2, 0.17.0, 0.17.1, 0.18.0, 0.18.1, 0.19.0, 0.19.1, 0.19.2, 0.20.0, 0.20.1, 0.20.2, 0.20.3, 0.21.0, 0.21.1, 0.22.0, 0.23.0, 0.23.1, 0.23.2, 0.23.3, 0.23.4, 0.24.0, 0.24.1, 0.24.2, 0.25.0, 0.25.1, 0.25.2, 0.25.3, 1.0.0, 1.0.1, 1.0.2, 1.0.3, 1.0.4, 1.0.5, 1.1.0, 1.1.1, 1.1.2, 1.1.3, 1.1.4, 1.1.5)
ERROR: No matching distribution found for pandas==1.3.4
clearml_agent: ERROR: Could not install task requirements!
while pandas==1.3.4 is easily available from pypi
SuccessfulKoala55 I've tried changing manually the TF version but it fails. I get:
import tensorflow as tf
File "/root/.clearml/venvs-builds/3.8/lib/python3.8/site-packages/tensorflow/init.py", line 435, in <module>
_ll.load_library(_main_dir)
File "/root/.clearml/venvs-builds/3.8/lib/python3.8/site-packages/tensorflow/python/framework/load_library.py", line 153, in load_library
py_tf.TF_LoadLibrary(lib)
tensorflow.python.framework.errors_impl.NotFoundError: /usr/local/lib/python3.8/dist-packages/tensorflow/core/kernels/libtfkernel_sobol_op.so: undefined symbol: _ZN10tensorflow14kernel_factory17OpKernelRegistrar12InitInternalEPKNS_9KernelDefEN4absl12lts_2021032411string_viewESt10unique_ptrINS0_15OpKernelFactoryESt14default_deleteIS9_EE
2021-11-14 18:09:10
Process failed, exit code 1
I assume every task you run on this container (latest Nvidia's TF container) will reproduce the same error upon importing of tensorflow
You should adjust the tensorflow version to match the image... Edit the task requirements for that
Well, assuming you have a ClearML Agent daemon running in docker mode, you should simply:
Clone your current task (right-click on the task in the tasks list and choose Clone) Go to the Execution tab Edit the Container section and set whatever docker image you need Enqueue the task to the queue watched by the daemon
SuccessfulKoala55 , meanwhile I try that, I encounter something weird. I am using a clearml agent with the following
clearml-agent daemon --detached --docker --gpus 0,1,2,3 --dynamic-gpus --queue kenny_1_gpu_queue=1
But for some reason although all the gpus are free and no other agent is on the machine, only one task is executed at the time instead of 4. Why is that?
I am also running from a NVIDIA container and I get
ERROR: No matching distribution found for tensorflow==2.4.0+nv
clearml_agent: ERROR: Could not install task requirements!
docker image is
http://nvcr.io/nvidia/tensorflow:21.10-tf2-py3
What should I do?
Well, on the first task it grabs it opens a different WORKER:gpu0 worker entry as expected while the agent stays with WORKER:dgpu0,1,2,3
but the other tasks on queue won't start and upon the first task's completion the following are not being run on WORKER:gpu0 but on WORKER:dgpu0,1,2,3 instead using only 1 GPU (the task execution says it runs on WORKER:gpu0)