#!/bin/bash
#
# GlusterFS and NFS Cluster Setup Script
# Configures a 3-node replica GlusterFS volume and mounts it + NFS across all 5 Swarm nodes.
#

set -euo pipefail

# Cluster Configuration
ALL_NODES=("140.44.4.71" "140.44.4.72" "140.44.4.73" "140.44.4.74" "140.44.4.75")
STORAGE_NODES=("140.44.4.71" "140.44.4.72" "140.44.4.73") # Nodes holding the actual data bricks

USER="snarf"
PASS="Katzir476!"

# NFS Configuration
NFS_SERVER="14.10.10.71"
NFS_SHARE="/volume1/docker" # Based on your nginxproxy.yml context

if ! command -v sshpass &> /dev/null; then
    echo "ERROR: 'sshpass' is required. Please install it first."
    exit 1
fi

# Helper function to run sudo commands remotely via sshpass
run_remote() {
    local node=$1
    local cmd=$2
    echo "   [${node}] Running command..."
    local escaped_cmd="${cmd//\'/\'\\\'\'}"
    sshpass -p "$PASS" ssh -o StrictHostKeyChecking=no "$USER@$node" "echo '$PASS' | sudo -S bash -c '$escaped_cmd'"
}

echo "=== Phase 1: Installing Prerequisites & Creating Directories ==="
for NODE in "${ALL_NODES[@]}"; do
    echo "-> Configuring Node $NODE"
    run_remote "$NODE" "
        apt-get update -y && \
        apt-get install -y glusterfs-server glusterfs-client nfs-common && \
        systemctl enable --now glusterd && \
        mkdir -p /data/glusterfs/swarm/brick && \
        mkdir -p /mnt/swarm_shared && \
        mkdir -p /mnt/nfs_shares
    "
done

echo ""
echo "=== Phase 2: Peering GlusterFS Storage Nodes ==="
PRIMARY_NODE="${STORAGE_NODES[0]}"

# Peer node 2 and 3 from node 1
for i in {1..2}; do
    PEER_NODE="${STORAGE_NODES[$i]}"
    echo "-> Peering $PRIMARY_NODE with $PEER_NODE"
    run_remote "$PRIMARY_NODE" "gluster peer probe $PEER_NODE"
done

# Give peering a few seconds to stabilize
sleep 5

echo ""
echo "=== Phase 3: Creating and Starting GlusterFS Volume ==="
echo "-> Creating 'replica 3' volume 'swarm_vols'..."
run_remote "$PRIMARY_NODE" "
    gluster volume create swarm_vols replica 3 \
    ${STORAGE_NODES[0]}:/data/glusterfs/swarm/brick \
    ${STORAGE_NODES[1]}:/data/glusterfs/swarm/brick \
    ${STORAGE_NODES[2]}:/data/glusterfs/swarm/brick \
    force || echo 'Volume might already exist.'
"

echo "-> Starting volume 'swarm_vols'..."
run_remote "$PRIMARY_NODE" "gluster volume start swarm_vols || echo 'Volume already started.'"

echo ""
echo "=== Phase 4: Mounting GlusterFS and NFS on ALL Nodes ==="
for NODE in "${ALL_NODES[@]}"; do
    echo "-> Mounting file systems on $NODE"
    run_remote "$NODE" "
        # Remove old fstab entries to prevent duplicates if script is re-run
        sed -i '/swarm_vols/d' /etc/fstab
        sed -i '/nfs_shares/d' /etc/fstab
        
        # Add to fstab
        # Using primary node IP with backup servers instead of localhost (required for client-only nodes)
        echo '140.44.4.71:/swarm_vols /mnt/swarm_shared glusterfs defaults,_netdev,backup-volfile-servers=140.44.4.72:140.44.4.73 0 0' >> /etc/fstab
        echo '$NFS_SERVER:$NFS_SHARE /mnt/nfs_shares nfs defaults,nfsvers=4,_netdev 0 0' >> /etc/fstab
        
        # Reload systemd so it recognizes the fstab changes
        systemctl daemon-reload
        
        # Mount them individually so one failure doesn't halt the other
        mount /mnt/swarm_shared
        mount /mnt/nfs_shares || echo 'NFS Mount Failed, continuing...'
    "
done

echo "=== Storage Cluster Setup Complete ==="