jmakov

Have the same problem: ```bash $ journalctl -u moneroocean_miner -f Apr 22 21:05:06 node1 systemd[1]: Stopped Monero miner service. Apr 22 21:05:06 node1 systemd[1]: Started Monero miner service. Apr 22...

Same problem here. Cannot use `ray` to distribute jobs on the cluster because `stattools.coint` is using all the cores on the driver node. Tried without success: ```python os.environ["OMP_NUM_THREADS"] = "1"...

Import order was the cause. Thank you. Would also be great to have this somewhere in the docs.

I now have the opposite problem - how to make sure all cores al used (for stattools.coint). ```python import os os.environ["MKL_NUM_THREADS"] = "24" os.environ["OPENBLAS_NUM_THREADS"] = "24" os.environ["OMP_NUM_THREADS"] = "24" from...

@bashtage thanks! But I'm interested why on ray cluster the driver node uses all the cores, but the worker nodes only 1 core. I'm not sure if it's a `statsmodels`...

For future reference when working with `ray`, this is the workaround: https://github.com/ray-project/ray/issues/25352.

Same problem here. @gdmsl thanks for sharing your solution! This also worked for me: `CMakeLists.txt`: ```cmake set(DEPENDENCIES_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../../lib) set(CSV_INCLUDE_DIR ${DEPENDENCIES_DIR}/csv-parser/include) include_directories(${CSV_INCLUDE_DIR}) ```

Is there any example how to run `ray` to compare models or train on a cluster?

Is there any update on this or is `ray` not planned to be supported?

Still an issue on 1.65.0