【发布时间】：2021-10-10 09:19:20
【问题描述】：
目标
在远程 MLflow 跟踪服务器中跟踪 R 模型（在 kubernetes 中运行）。该模型是在本地计算机上从 RStudio 开发的，在 Docker 容器中运行。
设置
根据我的研究，我需要创建安装了 conda 的 RStudio 映像。之后我想运行 MLflow 文档中的示例。
Dockerfile
FROM rocker/rstudio
USER root
ENV PATH="/root/miniconda3/bin:${PATH}"
ARG PATH="/root/miniconda3/bin:${PATH}"
ENV MLFLOW_BIN=/root/miniconda3/bin/mlflow
ENV MLFLOW_PYTHON_BIN=/root/miniconda3/bin/python
RUN apt-get update
RUN apt-get install -y wget && rm -rf /var/lib/apt/lists/*
RUN wget \
https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
&& mkdir /root/.conda \
&& bash Miniconda3-latest-Linux-x86_64.sh -b \
&& rm -f Miniconda3-latest-Linux-x86_64.sh
RUN conda --version
RUN R -e 'install.packages("mlflow")'
RUN R -e 'install.packages("glmnet")'
RUN R -e 'install.packages("carrier")'
RUN pip install -U mlflow==1.19.0
train.R（调整后的示例来自 here）
# The data set used in this example is from http://archive.ics.uci.edu/ml/datasets/Wine+Quality
# P. Cortez, A. Cerdeira, F. Almeida, T. Matos and J. Reis.
# Modeling wine preferences by data mining from physicochemical properties. In Decision Support Systems, Elsevier, 47(4):547-553, 2009.
library(mlflow)
library(glmnet)
library(carrier)
set.seed(40)
# Read the wine-quality csv file
data <- read.csv("wine-quality.csv")
# Split the data into training and test sets. (0.75, 0.25) split.
sampled <- sample(1:nrow(data), 0.75 * nrow(data))
train <- data[sampled, ]
test <- data[-sampled, ]
# The predicted column is "quality" which is a scalar from [3, 9]
train_x <- as.matrix(train[, !(names(train) == "quality")])
test_x <- as.matrix(test[, !(names(train) == "quality")])
train_y <- train[, "quality"]
test_y <- test[, "quality"]
alpha <- mlflow_param("alpha", 0.5, "numeric")
lambda <- mlflow_param("lambda", 0.5, "numeric")
Sys.setenv(MLFLOW_S3_ENDPOINT_URL="<EP>")
Sys.setenv(AWS_ACCESS_KEY_ID="<some_key>")
Sys.setenv(AWS_SECRET_ACCESS_KEY="<some_secret>")
mlflow_set_experiment("Wine R experiment")
mlflow_set_tracking_uri("<http...blabla>")
with(mlflow_start_run(), {
model <- glmnet(train_x, train_y, alpha = alpha, lambda = lambda, family= "gaussian", standardize = FALSE)
predictor <- crate(~ glmnet::predict.glmnet(!!model, as.matrix(.x)), !!model)
predicted <- predictor(test_x)
rmse <- sqrt(mean((predicted - test_y) ^ 2))
mae <- mean(abs(predicted - test_y))
r2 <- as.numeric(cor(predicted, test_y) ^ 2)
message("Elasticnet model (alpha=", alpha, ", lambda=", lambda, "):")
message(" RMSE: ", rmse)
message(" MAE: ", mae)
message(" R2: ", r2)
mlflow_log_param("alpha", alpha)
mlflow_log_param("lambda", lambda)
mlflow_log_metric("rmse", rmse)
mlflow_log_metric("r2", r2)
mlflow_log_metric("mae", mae)
mlflow_log_model(predictor, "model")
})
它的 POC 不介意不安全的环境变量。
我这样运行容器：
docker run --rm -p 8787:8787 -e PASSWORD=password --mount type=bind,source=$(pwd)/mlflow/examples/r_wine,target=/home/rstudio rstudio-mlflow
问题
每次从 RStudio 运行文件时，最后一行（mlflow_log_model(predictor, "model")）都会出错：
cannot start processx process '/root/miniconda3/bin/mlflow' (system error 13, Permission denied) @unix/processx.c:608 (processx_exec)
从 RStudio 终端列出 conda bin 文件夹时，我的权限被拒绝。您能帮我如何正确安装带有 RStudio 映像的 conda。
【问题讨论】：
-
我有点困惑 Conda 在这里的用途。如果要使用
pip安装所有东西，为什么不直接安装本机 Python？至于权限，Rocker下的RStudio以rstudio用户运行，没有提升权限，但Conda是root安装的。调整权限或安装为您计划运行的用户。 The Rocker docs 显示您可以在docker run中使用-e ROOT=TRUE向用户添加root 权限。 -
如果我正确理解了 mlflow 文档，它需要 conda env 来运行与 trackong 服务器的 coomucation（或者只是推荐）。我尝试了 env Root 但同样的问题。我可以在 docker -ti bash 之后将其作为 Rscrip train.r 运行，但不能从 RStudio 运行
标签： r docker rstudio conda mlflow