↓Skip to main content

Distributed Replicated Block Device (DRBD) - High-availability storage cluster

641 words·
DRBD High-availability Cluster
Table of Contents

Prerequisites
#

For the cluster nodes I use two Ubuntu 23.04 servers with the following IP addresses:

192.168.30.51 drbd1
192.168.30.52 drbd2

I use the following storage configuration on both nodes:

# List blockdevices
lsblk

# Shell output:
sda                         8:0    0   20G  0 disk
├─sda1                      8:1    0    1M  0 part
├─sda2                      8:2    0  1.8G  0 part /boot
└─sda3                      8:3    0 18.2G  0 part
  └─ubuntu--vg-ubuntu--lv 253:0    0   10G  0 lvm  /
sdb                         8:16   0    5G  0 disk

DRBD Setup
# 

# Install DRBD packages
sudo apt install drbd-utils -y

# Configure DNS in hosts file
sudo vi /etc/hosts

# Add cluster hostnames
192.168.30.51 drbd1
192.168.30.52 drbd2

DRBD Configuration
#

DRBD can be used on a hard drive, a partition, a RAID or an LVM logical volume. The size should be identical on both nodes.

# Configure DRBD: On both nodes
sudo vi /etc/drbd.conf

# /etc/drbd.conf
global { usage-count no; }
common { syncer { rate 100M; } }

resource drbd-storage { # Define name of DRBD storage resource
        protocol C; # Define protocol type
        startup {
                wfc-timeout  15;
                degr-wfc-timeout 60;
        }
        net {
                cram-hmac-alg sha1;
                shared-secret "secret";
        }

        on drbd1 { # Use hostname
                device /dev/drbd0; # Define mountpoint
                disk /dev/sdb; # Define storage device
                address 192.168.30.51:7788; # Define IP of node
                meta-disk internal;
        }
        on drbd2 { # Use hostname
                device /dev/drbd0; # Define mountpoint
                disk /dev/sdb; # Define storage device
                address 192.168.30.52:7788; # Define IP of node
                meta-disk internal;
        }
}

Protocol types

  • A Long distance replication
  • B Memory synchronous protocol
  • C Short distanced networks

Meta-Disk

  • internal Metadata is stored on same device as the data

Start DRBD
#

Initialize DRBD
# 

# Initialize the meta data storage: On both nodes
sudo drbdadm create-md drbd-storage
# or
sudo drbdadm create-md all

# Shell output:
initializing activity log
initializing bitmap (160 KB) to all zero
Writing meta data...
New drbd meta data block successfully created.

Start Daemon
# 

# Start the DRBD daemon: On both nodes
sudo systemctl start drbd.service

# Check blockdevices
lsblk

# Shell Output: Node 1
sda                         8:0    0    20G  0 disk
├─sda1                      8:1    0     1M  0 part
├─sda2                      8:2    0   1.8G  0 part /boot
└─sda3                      8:3    0  18.2G  0 part
  └─ubuntu--vg-ubuntu--lv 253:0    0    10G  0 lvm  /
sdb                         8:16   0     5G  0 disk
└─drbd1                   147:1    0     5G  1 disk

# Shell Output: Node 2
sda                         8:0    0    20G  0 disk
├─sda1                      8:1    0     1M  0 part
├─sda2                      8:2    0   1.8G  0 part /boot
└─sda3                      8:3    0  18.2G  0 part
  └─ubuntu--vg-ubuntu--lv 253:0    0    10G  0 lvm  /
sdb                         8:16   0     5G  0 disk
└─drbd0                   147:0    0     5G  1 disk

Start Synchronisation
# 

# Start the data synchronisation: Run on node 1 (define node 1 as primary node)
sudo drbdadm -- --overwrite-data-of-peer primary drbd-storage

# Watch synchronisation: Run on node 2
watch -n1 cat /proc/drbd

# Shell output: Wait till "cs:Connected"
Every 1.0s: cat /proc/drbd                drbd2: Thu Aug 31 18:42:30 2023

version: 8.4.11 (api:1/proto:86-101)
srcversion: 05D52F25C803AEC1A945DEF
 0: cs:SyncTarget ro:Secondary/Primary ds:Inconsistent/UpToDate C r-----
    ns:0 nr:1097728 dw:1097728 dr:0 al:8 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:f oos:4144956
        [===>................] sync'ed: 21.1% (4044/5116)M
        finish: 0:01:48 speed: 38,040 (37,852) want: 50,280 K/sec

# Check status
drbdadm status drbd-storage


# Shell output: Node 1
drbd-storage role:Primary
  disk:UpToDate
  peer role:Secondary
    replication:Established peer-disk:UpToDate

# Shell output: Node 2
drbd-storage role:Secondary
  disk:UpToDate
  peer role:Primary
    replication:Established peer-disk:UpToDate

Create Filesystem
# 

# Create ext4 filesystem: Run on node 1
sudo mkfs.ext4 /dev/drbd0

# Create mountpoint: Run on node 1
sudo mkdir -p /mnt/drbd-storage

# Mount DRDB storage: Run on node 1
sudo mount /dev/drbd0 /mnt/drbd-storage

Switch DRBD Roles
# 

# Unmount DRBD resource
sudo umount /mnt/drbd-storage

# Switch node 1 to secondary role
sudo drbdadm secondary drbd-storage

# Switch node 2 to primary role
sudo drbdadm primary drbd-storage

# Create mountpoint: Run on node 2
sudo mkdir -p /mnt/drbd-storage

# Mount DRDB storage: Run on node 2
sudo mount /dev/drbd0 /mnt/drbd-storage