RLBackfilling Using PyTorch

Important

We noticed that the way we calculated the bsld is not accurate after the paper was published. Hence, we corrected it in this update (bad9bc3) and also fixed the corresponding tables/plots in the paper on Arxiv

This repo includes the bfTorch source code and necessary datasets to run the experiments/tests.

The code has been tested on Ubuntu 18.04 with PyTorch 1.13 and Gym 0.21.

Installation

Required Software

• Python 3.9 and PyTorch Use VirtuanEnv or Conda to build a Python3.9 environment and PyTorch at least 1.13.0 Note that, we do not leverage GPUs, so no need to configure the GPU version of PyTorch.

• OpenMPI and mpi4py

sudo apt-get install openmpi-bin openmpi-doc libopenmpi-dev
conda install mpi4py

Clone RLBackfilling

git clone https://github.com/DIR-LAB/RLBackfilling.git

Install Dependencies

cd RLBackfilling
pip install -r requirements.txt

File Structure

data/: Contains a series of workload and real-world traces.
plot.py: Plot the trained results.
rlbackfill.py: Used to train and run RLBackfilling models, as well as tests for actual/noisy runtime performance
rlbackfill-test.py: Generates raw avgbsld scores and box+whisker plot.

To change the hyper-parameters, such as MAX_OBSV_SIZE or the trajectory length during training, you can change them in rlbackfill.py.

Training

To train a RL model based on a job trace, run this command:

python rlbackfill.py --workload ./data/lublin_256.swf --exp_name your-exp-name --trajs 400 --epochs 300 --heuristic fcfs

There are many other parameters in the source file.

• --model, specify a saved trained model (for two-step training and re-training)
• --pre_trained, specify whether this trainig will be a twp-step training or re-training

Monitor Training

After running Default Training, a folder named logs/your-exp-name/ will be generated.

python plot.py ./data/logs/your-exp-name/ -x Epoch -s 1

It will plot the training curve.

Test and Compare

After the RLBackfiller converges, you can test the result and compare it with different policies such as FCFS, SJF, WFP3, UNICEP, and F1.

python rlbackfill-test.py --rlmodel "./logs/your-exp-name/your-exp-name_s0/" --workload "./data/lublin_256.swf" --len 2048 --iter 10

There are many parameters you can use:

• --seed, the seed for random sampling
• --iter, how many iterations for the testing

Runtime Testing

To test the how differences in runtime accuracy can affect results, modify the self.request_time in the Job class located in rlbackfill.py For example, to test how 100% accurate request time affects scheduling, perform this change in the code

self.request_time = self.run_time