DRBD

DRBD is the service used for synchronisation of data (usually web and database files) on a cluster or HA solution. Not to be confused with Unision (which is typically used for synchronisation of config files due to not updating files as quickly as DRBD).

Installing

yum install drbd-utils* drbd-dkms

apt-get install drbd-utils* drbd-dkms

may need to add repos for this.

The above installs both the DRBD service files, as well as the required kernel module.

Check that the kernel module is loaded with:

modprobe drbd

Configuration of disks for DRBD

https://clusterlabs.org/pacemaker/doc/2.1/Clusters_from_Scratch/epub/shared-storage.html

DRBD will need its own block device on each node.

In this example, I've added a 10GB disk to each node,

root@b4sed-02:/etc# lsblk
NAME            MAJ:MIN RM  SIZE RO TYPE MOUNTPOINTS
sda               8:0    0 19.1G  0 disk
├─sda1            8:1    0 18.8G  0 part /
├─sda14           8:14   0    1M  0 part
└─sda15           8:15   0  256M  0 part /boot/efi
sdb               8:16   0   10G  0 disk
└─VG_data-DRBD2 253:0    0   10G  0 lvm
sr0              11:0    1 1024M  0 rom


root@b4sed-02:/home# pvcreate /dev/sdb
  Physical volume "/dev/sdb" successfully created.


root@b4sed-02:/home# vgcreate VG_data /dev/sdb
  Volume group "VG_data" successfully created


root@b4sed-02:/home# vgdisplay | grep -e Name -e Free
  VG Name               VG_data
  Free  PE / Size       2559 / <10.00 GiB


root@b4sed-02:/home# lvcreate --name DRBD2 -l2559 VG_data
  Logical volume "DRBD2" created.

Configuration of DRBD

typically configured via /etc/drbd.conf

example configuration from my setup:

resource wwwdata {
 protocol C;
 meta-disk internal;
 device /dev/drbd1;
 syncer {
  verify-alg sha1;
 }
 net {
  allow-two-primaries;
 }
 on b4sed-01 {
  disk   /dev/VG_data/DRBD1;
  address  10.0.0.2:7789;
 }
 on b4sed-02 {
  disk   /dev/VG_data/DRBD2;
  address  10.0.0.3:7789;
 }
}

Once configuration file is in place, DRBD can be deployed via the below commands:

drbdadm create-md resourcename #wwwdata in my example

modprobe drbd

drbdadm up resourcename

as a side note, in my example I was getting errors relating to the kernel module, a kernel update and reboot resolved this.

Check status

drbdadm status

cat /proc/drbd

At this point you'll likely see data inconsistency:

root@b4sed-01:/var/log# drbdadm status
wwwdata role:Secondary
  disk:Inconsistent
  b4sed-02 connection:Connecting

This is because the data might differ on each node, to specific which node has the correct data we need to set the primary node, using the below command:

drbdadm primary --force wwwdata

Once done, run the deployment commands again but on the 2nd node:

drbdadm create-md resourcename #wwwdata in my example

modprobe drbd

drbdadm up resourcename

Once done, give some time for the connection to be made, check the status again:

root@b4sed-01:~# drbdadm status
wwwdata role:Primary
  disk:UpToDate
  b4sed-02 role:Secondary
    peer-disk:UpToDate

Configure the DRBD disk

mkfs.ext4 /dev/drbd1

nearly there now...

Add DRBD to the cluster

pcs cluster cib drbd_cfg #this queues up changed to be deployed to the cluster in one go

add the constraints and such:

pcs -f fs_cfg resource create WebFS Filesystem \
    device="/dev/drbd1" directory="/var/www/html" fstype="ext4"

pcs -f fs_cfg constraint colocation add \
    WebFS with WebData-clone INFINITY with-rsc-role=Master

pcs -f fs_cfg constraint order \
    promote WebData-clone then start WebFS

pcs -f fs_cfg constraint colocation add http_server with WebFS INFINITY

pcs -f fs_cfg constraint order WebFS then http_server

if all looks good, you can push these changes with the below command:

pcs cluster cib-push fs_cfg --config

time now for testing - place 1 node in standby and ensure the failover is sucesful.