Merge slurm sim lite into this repo

docker-compose.yml → docker/docker-compose.yml

@@ -1,6 +1,6 @@
 services:
   slurm-mira:
-    image: dvalts/slurm-sim-cirrus
+    image: decvalts/slurm_sim_lite
     container_name: slurmsim
     hostname: slurmsim
     build:

docker/slurm_sim_lite/Dockerfile

@@ -0,0 +1,132 @@
+FROM centos:7
+
+LABEL desc="Slurm simulator made ready"
+
+#Adding in the code from github to be able to start/stop mysql (and sshd?)
+COPY cmd_start /sbin/
+COPY cmd_stop /sbin/
+
+
+# giving permissions to use the cmd from above
+RUN \
+	chmod a+rwx /sbin/cmd_start && \
+	chmod a+rwx /sbin/cmd_stop && \
+        mkdir /install_files && \
+        useradd -d /home/slurm -ms /bin/bash slurm && \
+        usermod -aG wheel slurm && \
+        echo "slurm:slurm"|chpasswd && \
+        echo "Added slurm user" && \
+	yum -y install git && \
+	yum clean all
+
+# getting file that installs all the R packages
+COPY ./package_install.R /install_files
+
+# creating all the directories needed for larger run command
+USER slurm 
+RUN \
+        cd /home/slurm && \
+        mkdir slurm_sim_ws && \
+        cd slurm_sim_ws && \
+        mkdir sim && \
+        cd /home/slurm/slurm_sim_ws && \
+        git clone https://github.com/ubccr-slurm-simulator/slurm_sim_tools.git
+
+USER root
+
+
+# installing mysql (mariadb), python, and R, setting everything up all in one Run command 
+RUN \
+	yum -y install mariadb-server && \
+	yum -y install mariadb-devel && \
+	echo "Done installing Mariadb" && \
+	yum -y install gcc-c++ && \
+	yum -y install install epel-release && \
+	yum -y install python36 python36-libs python36-devel python36-numpy python36-scipy python36-pip
+
+RUN \
+	pip3 install pymysql && \
+	pip3 install pandas && \
+	echo "Python all installed" && \
+	yum -y install R R-Rcpp R-Rcpp-devel && \
+	yum -y install python-devel && \
+	yum -y install texlive-* && \
+	echo "R all installed" && \
+	Rscript /install_files/package_install.R && \
+        echo "Installed R packages" && \
+        yum -y install sudo && \
+        yum -y install wget && \
+	echo "Sudo, git, wget installed" && \
+        wget https://download2.rstudio.org/server/centos6/x86_64/rstudio-server-rhel-1.2.5042-x86_64.rpm && \
+        sudo yum -y install rstudio-server-rhel-1.2.5042-x86_64.rpm && \
+        yum -y install initscripts && \
+	echo "Rstudio server installed" && \
+        yum -y install openssh openssh-server openssh-clients openssl-libs && \
+        mkdir /var/run/sshd && \
+        ssh-keygen -t rsa -f /etc/ssh/ssh_host_rsa_key -N '' && \
+        ssh-keygen -t ecdsa -f /etc/ssh/ssh_host_ecdsa_key -N '' && \
+        ssh-keygen -t ed25519 -f /etc/ssh/ssh_host_ed25519_key -N '' && \
+	echo "Ssh installed" && \
+        chmod g+rw /var/lib/mysql /var/log/mariadb /var/run/mariadb && \
+        mysql_install_db && \
+        chown -R mysql:mysql /var/lib/mysql && \
+        cmd_start mysqld && \
+        mysql -e "create user 'slurm'@'localhost' identified by 'slurm';" && \
+        mysql -e "GRANT ALL PRIVILEGES ON *.* TO 'slurm'@'localhost' IDENTIFIED BY 'slurm';" && \
+        cmd_stop mysqld && \
+	yum clean all
+
+
+# switch to slurm user so the next directories made are owned by slurm
+USER slurm 
+
+# installing slurm simulator
+RUN \
+	cd /home/slurm/slurm_sim_ws && \
+        git clone https://github.com/ubccr-slurm-simulator/slurm_simulator.git && \
+        cd slurm_simulator && \
+        cd .. && \
+        mkdir bld_opt && \
+        cd bld_opt && \
+        ../slurm_simulator/configure --prefix=/home/slurm/slurm_sim_ws/slurm_opt --enable-simulator \
+--enable-pam --without-munge --enable-front-end --with-mysql-config=/usr/bin/ --disable-debug \
+CFLAGS="-g -O3 -D NDEBUG=1" && \
+        make -j install
+
+
+# 8787 is the default port that rstudio server uses, so need to expose it to use it
+EXPOSE 8787
+
+
+USER root
+
+
+COPY ./startup_file.sh /install_files
+COPY ./initial_test.sh /install_files
+COPY ./micro_cluster_setup.py /install_files
+COPY ./micro_ws_config.sh /install_files
+COPY ./populate_slurmdb.sh /install_files
+COPY ./generate_job_trace.sh /install_files
+COPY ./run_sim.sh /install_files
+COPY ./check_results.R /install_files
+
+
+# need to expose port 22 to allow for ssh to work properly
+EXPOSE 22 
+
+
+# expose for mysql use
+EXPOSE 3306
+
+
+# back to root for easier permissions stuff
+RUN \
+	chmod -R a+rwx /install_files
+
+
+# sets cmd_start as entrypoint, then runs the startup file and the initial test file
+ENTRYPOINT ["/sbin/cmd_start"]
+CMD ["/install_files/startup_file.sh","/install_files/initial_test.sh"]
+
+
+

docker/slurm_sim_lite/README.md

@@ -0,0 +1,84 @@
+# Slurm Simulator Docker and Apptainer Containers
+
+```bash
+docker build -f docker/slurm_sim/slurm_sim.Dockerfile -t nsimakov/slurm_sim:v3.0 .
+
+# debug mode
+# older docker uses --invoke
+# BUILDX_EXPERIMENTAL=1 docker buildx build --invoke /bin/bash -f docker/slurm_sim/slurm_sim.Dockerfile -t nsimakov/slurm_sim:v3.0 .
+BUILDX_EXPERIMENTAL=1 docker buildx debug build -f docker/slurm_sim/slurm_sim.Dockerfile -t nsimakov/slurm_sim:v3.0 .
+
+# run
+docker run -p 0.0.0.0:8888:8888 -it --rm \
+    --name slurmsim -h slurmsim \
+    -v $PWD/tutorials:/home/jovyan/work -v $PWD:/opt/slurm_sim_tools \
+    -v /tmp/.X11-unix:/tmp/.X11-unix -v /mnt/wslg:/mnt/wslg \
+    -e DISPLAY=$DISPLAY -e WAYLAND_DISPLAY=$WAYLAND_DISPLAY \
+    -e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR -e PULSE_SERVER=$PULSE_SERVER \
+    nsimakov/slurm_sim:v3.0
+
+docker run -it -v /tmp/.X11-unix:/tmp/.X11-unix  \
+    -e DISPLAY=$DISPLAY -e WAYLAND_DISPLAY=$WAYLAND_DISPLAY \
+    -e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR -e PULSE_SERVER=$PULSE_SERVER xclock
+
+docker run -p 0.0.0.0:8888:8888 -it --rm \
+    --name slurmsim -h slurmsim \
+    -v $PWD/tutorials:/home/jovyan/work -v $PWD:/opt/slurm_sim_tools \
+    -v /tmp/.X11-unix:/tmp/.X11-unix -v /mnt/wslg:/mnt/wslg \
+    -e DISPLAY=$DISPLAY -e WAYLAND_DISPLAY=$WAYLAND_DISPLAY \
+    -e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR -e PULSE_SERVER=$PULSE_SERVER \
+    -e NB_USER="slurm" \
+    -e NB_UID="1000" \
+    -e NB_GROUP="slurm" \
+    -e NB_GID="1000" \
+    -e CHOWN_HOME=yes \
+    nsimakov/slurm_sim:v3.0 bash
+
+
+```
+Done running hooks in: /usr/local/bin/start-notebook.d
+Update jovyan's UID:GID to 1000:100
+Running hooks in: /usr/local/bin/before-notebook.d as uid: 0 gid: 0
+Done running hooks in: /usr/local/bin/before-notebook.d
+
+
+in container run following to set time:
+
+```
+Run 'dpkg-reconfigure tzdata' if you wish to change it.
+```
+
+# Slurm Simulator Docker and Singularity Containers v1.2
+
+```bash
+docker run -p 0.0.0.0:8888:8888 -it --rm -h slurmsim nsimakov/slurm_sim:v3.0
+
+```
+
+Building
+## from repo root
+docker build -f ./docker/slurm_sim/Dockerfile -t nsimakov/ub-slurm-sim:v1.2 .
+docker push nsimakov/ub-slurm-sim:v1.2
+Running
+
+
+## Singularity Installation
+
+```bash
+export VERSION=1.18 OS=linux ARCH=amd64 && \  # Replace the values as needed
+wget https://dl.google.com/go/go$VERSION.$OS-$ARCH.tar.gz && \ # Downloads the required Go package
+sudo tar -C /usr/local -xzvf go$VERSION.$OS-$ARCH.tar.gz && \ # Extracts the archive
+rm go$VERSION.$OS-$ARCH.tar.gz    # Deletes the ``tar`` file
+
+echo 'export PATH=/usr/local/go/bin:$PATH' >> ~/.bashrc && \
+source ~/.bashrc
+
+export VERSION=3.9.7 && # adjust this as necessary \
+    wget https://github.com/sylabs/singularity/releases/download/v${VERSION}/singularity-ce-${VERSION}.tar.gz && \
+    tar -xzf singularity-ce-${VERSION}.tar.gz && \
+    cd singularity-ce-${VERSION}
+./mconfig && \
+    make -C builddir && \
+    sudo make -C builddir install
+
+```

docker/slurm_sim_lite/check_results.R

@@ -0,0 +1,71 @@
+#!/usr/bin/env Rscript
+
+# This script gets the results from the simulation and runs some tests on them 
+# It tests if the requested features were given to the jobs
+# Features: cpu type, gpu, big mem
+# How implemented - each feature corresponds to a different type of node
+
+
+library(RSlurmSimTools) # needs these libraries to run the tests
+library(dplyr)
+
+# this function allows easier comparison to see if the simulator assigned things correctly
+# checks the trace value (requested feature) against the sacct value (Node List, indicating assigning of a feature)
+# the check values are there for reuse of the function for more than one type of test
+check_nodes <- function(df.joined, row_num, trace_col, trace_check, sacct_col, sacct_check){
+	result = TRUE # assumes correct
+	# df.joined is the joined data frame from trace and sacct data frames
+	trace_val = df.joined[row_num, trace_col] # trace value (feature)
+	sacct_val = df.joined[row_num, sacct_col] # sim value (if implemented feature)
+
+	# no feature requested if the value is NA, so check for that
+	if(!is.na(trace_val)) 
+	{
+		# check if the feature (trace_check) was requested
+		if(trace_val == trace_check) 
+		{
+			# checks if the node list has the node corresponding to that feature
+			if(!(grepl(sacct_check, sacct_val))) 
+			{
+				# if improper nodes have been assigned, its a false result (didn't assign properly)
+				result = FALSE 
+			}
+		}
+	}
+	result # result is returned	
+}
+
+
+# reads in the csv file of the job traces (jobs submitted)
+job_trace <- read.csv(file="/home/slurm/slurm_sim_ws/slurm_sim_tools/reg_testing/micro_cluster/test_trace.csv")
+
+# reads in log file of resulting data (what jobs were assigned, where, etc)
+sacct_base <- read_sacct_out("/home/slurm/slurm_sim_ws/sim/micro/baseline/results/jobcomp.log")
+
+# creating a joined data frame by job id so that can go through jobs easier
+joined <- left_join(job_trace, sacct_base, by = c("sim_job_id" = "local_job_id") )
+
+done_well = TRUE # assumes did correctly
+
+# loops through each row in the joined data frame
+for(row in 1:nrow(joined))
+{
+	# checks if all features have been met (or weren't present)
+    done_well = check_nodes(joined, row, "sim_req_mem", 500000, "NodeList", "b") && # big mem
+      check_nodes(joined, row, "sim_features", "CPU-M", "NodeList", "m") && # M cpu
+      check_nodes(joined, row, "sim_features", "CPU-N", "NodeList", "n") && # N cpu
+      check_nodes(joined, row, "sim_gres", "gpu:1", "NodeList", "g") && # 1 gpu
+      check_nodes(joined, row, "sim_gres", "gpu:2", "NodeList", "g") # 2 gpu
+
+	# if at any point a feature doesn't match, breaks out of the loop
+	if(!done_well) 
+	{
+		# prints out the job id for tracing back what failed
+		jobid = joined[row, "sim_job_id"]
+		print(paste("Id of incorrectly assigned job:", jobid))
+		break
+	}
+}
+# prints overall result
+print("Did the simulator do well?.....")
+print(done_well)