fix
This commit is contained in:
parent
960ca08b09
commit
47c72eec7e
17
Dockerfile
17
Dockerfile
@ -1,6 +1,6 @@
|
||||
FROM debian:stable-slim
|
||||
|
||||
# Install g++, make, OpenMPI, AND the OpenSSH server
|
||||
# ... (all the apt-get install and ssh setup lines are unchanged) ...
|
||||
RUN apt-get update && apt-get install -y --no-install-recommends \
|
||||
g++ \
|
||||
make \
|
||||
@ -9,28 +9,17 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
|
||||
openssh-server && \
|
||||
rm -rf /var/lib/apt/lists/*
|
||||
|
||||
# Create a directory for the SSH daemon to run (use -p to avoid errors)
|
||||
RUN mkdir -p /var/run/sshd
|
||||
|
||||
# --- SSH Key Generation and Configuration ---
|
||||
# Generate a passwordless SSH key pair
|
||||
RUN ssh-keygen -t rsa -f /root/.ssh/id_rsa -N ''
|
||||
|
||||
# Add the public key to the list of authorized keys for passwordless login
|
||||
RUN cat /root/.ssh/id_rsa.pub >> /root/.ssh/authorized_keys
|
||||
RUN chmod 600 /root/.ssh/authorized_keys
|
||||
|
||||
# --- THIS IS THE NEW LINE ---
|
||||
# Create an SSH config file to disable host key checking
|
||||
RUN echo "Host *\n\tStrictHostKeyChecking no" >> /root/.ssh/config
|
||||
# --- END OF NEW LINE ---
|
||||
|
||||
# Allow root login via SSH
|
||||
RUN sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config
|
||||
|
||||
WORKDIR /app
|
||||
|
||||
COPY . /app/
|
||||
|
||||
# Compile using the MPI C++ compiler wrapper
|
||||
# Compile ALL the necessary tasks during the build
|
||||
RUN mpic++ -O2 -std=c++17 -o mpi_task task.cpp
|
||||
RUN mpic++ -O2 -std=c++17 -o task1 task1.cpp
|
||||
@ -2,16 +2,18 @@ services:
|
||||
master:
|
||||
build: .
|
||||
container_name: pdc-master
|
||||
hostname: pdc-master # <-- ADD THIS LINE
|
||||
volumes:
|
||||
- ./hostfile:/app/hostfile
|
||||
command: >
|
||||
sh -c "/usr/sbin/sshd & sleep 5 && mpirun -np 4 --hostfile /app/hostfile --allow-run-as-root /app/mpi_task"
|
||||
sh -c "/usr/sbin/sshd & sleep 5 && mpirun -np 5 --hostfile /app/hostfile --allow-run-as-root /app/task1"
|
||||
networks:
|
||||
- pdcn
|
||||
|
||||
worker1:
|
||||
build: .
|
||||
container_name: pdc-worker1
|
||||
hostname: pdc-worker1 # <-- ADD THIS LINE
|
||||
command: /usr/sbin/sshd -D
|
||||
networks:
|
||||
- pdcn
|
||||
@ -19,6 +21,7 @@ services:
|
||||
worker2:
|
||||
build: .
|
||||
container_name: pdc-worker2
|
||||
hostname: pdc-worker2 # <-- ADD THIS LINE
|
||||
command: /usr/sbin/sshd -D
|
||||
networks:
|
||||
- pdcn
|
||||
@ -26,6 +29,15 @@ services:
|
||||
worker3:
|
||||
build: .
|
||||
container_name: pdc-worker3
|
||||
hostname: pdc-worker3 # <-- ADD THIS LINE
|
||||
command: /usr/sbin/sshd -D
|
||||
networks:
|
||||
- pdcn
|
||||
|
||||
worker4:
|
||||
build: .
|
||||
container_name: pdc-worker4
|
||||
hostname: pdc-worker4 # <-- ADD THIS LINE
|
||||
command: /usr/sbin/sshd -D
|
||||
networks:
|
||||
- pdcn
|
||||
|
||||
1
hostfile
1
hostfile
@ -2,3 +2,4 @@ master slots=1
|
||||
worker1 slots=1
|
||||
worker2 slots=1
|
||||
worker3 slots=1
|
||||
worker4 slots=1
|
||||
78
task1.cpp
Normal file
78
task1.cpp
Normal file
@ -0,0 +1,78 @@
|
||||
#include <iostream>
|
||||
#include <vector>
|
||||
#include <mpi.h>
|
||||
#include <unistd.h> // Required for gethostname() and getpid()
|
||||
|
||||
int main(int argc, char** argv) {
|
||||
MPI_Init(&argc, &argv);
|
||||
|
||||
int world_size;
|
||||
MPI_Comm_size(MPI_COMM_WORLD, &world_size);
|
||||
|
||||
int world_rank;
|
||||
MPI_Comm_rank(MPI_COMM_WORLD, &world_rank);
|
||||
|
||||
// Get the hostname of the container/machine the process is running on.
|
||||
char hostname[256];
|
||||
gethostname(hostname, sizeof(hostname));
|
||||
|
||||
// Get the Process ID of this specific MPI process.
|
||||
pid_t pid = getpid();
|
||||
|
||||
int num_workers = world_size - 1;
|
||||
|
||||
if (world_rank == 0) {
|
||||
// --- MASTER PROCESS ---
|
||||
int data[] = {1, 2, 3, 4};
|
||||
int data_count = sizeof(data) / sizeof(int);
|
||||
|
||||
if (data_count != num_workers) {
|
||||
if (world_rank == 0) { // Only master should print the error
|
||||
std::cerr << "Error: The number of workers (" << num_workers
|
||||
<< ") must match the number of data elements (" << data_count << ").\n"
|
||||
<< "Please run with -np " << data_count + 1 << " processes." << std::endl;
|
||||
}
|
||||
MPI_Abort(MPI_COMM_WORLD, 1);
|
||||
}
|
||||
|
||||
// Updated print statement to include PID and Hostname
|
||||
std::cout << "Master (rank 0, PID " << pid << " on " << hostname << "): sending numbers 1,2,3,4 to workers..." << std::endl;
|
||||
|
||||
for (int i = 0; i < num_workers; ++i) {
|
||||
int worker_rank = i + 1;
|
||||
MPI_Send(&data[i], 1, MPI_INT, worker_rank, 0, MPI_COMM_WORLD);
|
||||
}
|
||||
|
||||
std::vector<int> results(num_workers);
|
||||
for (int i = 0; i < num_workers; ++i) {
|
||||
int worker_rank = i + 1;
|
||||
int received_result;
|
||||
MPI_Recv(&received_result, 1, MPI_INT, worker_rank, 0, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
|
||||
results[i] = received_result;
|
||||
}
|
||||
|
||||
// Updated print statement
|
||||
std::cout << "Master (rank 0, PID " << pid << " on " << hostname << "): collected results {";
|
||||
for (int i = 0; i < results.size(); ++i) {
|
||||
std::cout << results[i] << (i == results.size() - 1 ? "" : ",");
|
||||
}
|
||||
std::cout << "}" << std::endl;
|
||||
|
||||
} else {
|
||||
// --- WORKER PROCESSES ---
|
||||
int received_number;
|
||||
MPI_Recv(&received_number, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
|
||||
|
||||
int computed_result = received_number * 10;
|
||||
|
||||
// Updated print statement to include PID and Hostname
|
||||
std::cout << "Worker (rank " << world_rank << ", PID " << pid << " on " << hostname
|
||||
<< "): received " << received_number
|
||||
<< ", computed " << computed_result << std::endl;
|
||||
|
||||
MPI_Send(&computed_result, 1, MPI_INT, 0, 0, MPI_COMM_WORLD);
|
||||
}
|
||||
|
||||
MPI_Finalize();
|
||||
return 0;
|
||||
}
|
||||
Loading…
Reference in New Issue
Block a user