Archives For storage

DRBD (Distributed Redundant Block Device) is an open source storage solution that is best compared with a RAID-1 (mirror) between two servers. I’ve implemented this for both our cloud storage as our cloud management servers.

We’re in the process of replacing both cloud storage nodes (that is: everything except the disks and its raid array) and of course no downtime is allowed. Although DRBD is made for redundancy (one node can be offline without impact), completely replacing a node is a bit tricky.

Preventing a ‘split brain’ situation
The most important thing to remember is that only one storage node is allowed to be the active node at all times. If this is violated, a so-called ‘split brain’ happens. DRBD has methods in surviving such a state, but it is best to prevent it from happening.

When discussing this project in our Team, it was suggested to boot the replaced storage node without any networking cables. Usually this is a safe way to prevent the node from interacting with others. In this case, it is not such a good idea: since the new secondary server has the same disks and configuration as the old one unexpeced thing may happen. When booting without networking cables attached, both nodes cannot find one another and the newly booted secondary may decode to become primary itself. A split brain situation will then occur: both nodes will be primary at the same time and access the data. You won’t be able to recover from this, unless you manually decide which node is the master (and lose the changes on the other node). In this case this can be easily decided, but it’s a lot of unneccessary trouble.

Instead, boot the replaced node with network cables connected so the replication network will be up and both nodes immediately will see each other. In our case, this means connecting the 10Gbps connection between the nodes. This connection is used by DRBD for syncing and by Heartbeat for sending the heartbeats. This prevents entering the ‘split brain’ state and immediately starts syncing.

Note: If you want to replace everything including the disks, you’ll have to manually join the cluster with the new secondary node and then sync the data. In this case it doesn’t matter whether the networking cables are connected or not, since this new node won’t be able to become primary anyway.

The procedure
Back to our case: replacing all hardware except for the disks. We managed to successfully replace both nodes using this procedure:

  1. shut down the secondary node
  2. replace the hardware, install the existing disks and 10Gbps card
  3. boot the node with at least the 10Gbps connection active
  4. the node should sync with the primary
  5. when syncing finishes, redundancy is restored
  6. make sure all other networking connections are working. Since the main board was replaced some MAC-addresses changes. Update UDEV accordingly
  7. when all is fine, check if DRBD and Heartbeat are running without errors on both nodes
  8. then stop heartbeat on the primary node. A fail-over to the new secondary node will occur
  9. if all went well, you can now safely shut down the old primary
  10. replace the hardware, install the existing disks and 10Gbps card
  11. boot the node with at least the 10Gbps connection active
  12. the node should sync with the primary
  13. when syncing finishes, redundancy is restored
  14. make sure all other networking connections are working. Since the main board was replaced some MAC-addresses changes. Update UDEV accordingly
  15. the old primary is now secondary
  16. If you want, initialize another fail-over (In our case we didn’t fail-over again, since both nodes are equal powerful)

Congratulations: the cluster is redundant again with the new hardware!

Using the above procedure, we replaced both nodes of our DRBD storage cluster without any downtime.

A few days ago we installed a ‘Battery Backup Unit’ in our secondary storage server. This allows us to turn on the ‘Write Back Cache’. The performance impact was impressive..

Enabling the Write Back Cache means writes are committed to the raid controler’s cache (which is much faster) so you don’t have to wait for the data to be written to disk. Normally, this is a risky operation because when the power goes down unexpectedly the data in the raid controller’s cache is lost. Thanks to the battery the raid controller can finish all of its writes to the disk even there is no more power.

Have a look at the below graph. It shows the load dropped significantly after we’ve installed this battery.

Starting on the left you see normal operations until we switched off the server around midnight. All services kept working by the way, but more about that redundancy magic in another post. The big spike around 1am was caused by syncing the data with the primary storage again after the server came online again. We had not turned on the Write Back Cache at that time. When it finished syncing, we rebooted the server once again, upgraded firmware and activated the Write Back Cache. We immediately saw an performance boost of around 20 times! The small spike around 2am was syncing with the primary storage again, but this time with Write Back Cache enabled. Our load averages now peak at 1 or 2 instead of >20.

Lesson learned: always install a Battery Backup Unit so you can safely turn the Write Back Cache on!

I’ve been building redundant storage solutions for years. At first, I used it for our webcluster storage. Nowadays it’s the base of our CloudStack Cloud-storage. If you ask me, the best way to create a redundant pair of Linux storage servers using Open Source software, is to use DRBD. Over the years it has proven to be rock solid to me.

DRBD is a Distributed Replicated Block Device. You can think of DRBD as RAID-1 between two servers. Data is mirrored from the primary to the secondary server. When the primary fails, the secondary takes over and all services remain online. DRBD provides tools for failover but it does not handled the actual failover. Cluster management software like Heartbeat and PaceMaker are made for this.

In this post I’ll show you how to install and configure DRBD, create file systems using LVM2 on top of the DRBD device, serve the file systems using NFS and manage the cluster using Heartbeat.

Installing and configuring DRBD
I’m using mostly Debian so I’ll focus on this OS. I did setup DRBD on CentOS as well. You need to use the ELREPO repository to find the right packages.

Squeeze-backports has a newer version of DRBD. If you, like me, want to use this version instead of the one in Squeeze itself, use this method to do so:

echo "
deb squeeze-backports main contrib non-free
" >> /etc/apt/sources.list

echo "Package: drbd8-utils
Pin: release n=squeeze-backports
Pin-Priority: 900
" > /etc/apt/preferences.d/drbd

Then install the DRBD utils:

apt-get update
apt-get install drbd8-utils

As the DRBD-servers work closely together, it’s important to keep the time synchronised. Install a NTP system for this job.

apt-get install ntp ntpdate

You also need a kernel module but that one is in the stock Debian kernel. If you’re compiling kernels yourself, make sure to include this module. When you’re ready, load the module:

modprobe drbd

Verify if all went well by checking the active modules:

lsmod | grep drbd

The expected output is something like:

drbd 191530 4 
lru_cache 12880 1 drbd
cn 12933 1 drbd

Most online tutorials instruct you to edit ‘/etc/drbd.conf’. I’d suggest not to touch that file and create one in /etc/drbd.d/ instead. That way, your changes are never overwritten and it’s clear what local changed you made.

vim /etc/drbd.d/redundantstorage.res

Enter this configuration:

resource redundantstorage {
 protocol C;
 startup { wfc-timeout 0; degr-wfc-timeout 120; }

disk { on-io-error detach; }
 on {
  device /dev/drbd0;
  disk /dev/sda3;
  meta-disk internal;
 on {
  device /dev/drbd0;
  disk /dev/sda3;
  meta-disk internal;

Make sure your hostnames match the hostnames in this config file as it will not work otherwise. To see the current hostname, run:

uname -n

Modify /etc/hosts, /etc/resolv.conf and/or /etc/hostname to your needs and do not continue until the actual hostname matches the one you set in the configuration above.

Also, make sure you did all the steps so far on both servers.

It’s now time to initialise the DRBD device:

drbdadm create-md redundantstorage
drbdadm up redundantstorage
drbdadm attach redundantstorage
drbdadm syncer redundantstorage
drbdadm connect redundantstorage

Run this on the primary server only:

drbdadm -- --overwrite-data-of-peer primary redundantstorage

Monitor the progress:

cat /proc/drbd

Start the DRBD service on both servers:

service drbd start

You now have a raw block device on /dev/drbd0 that is synced from the primary to the secondary server.

Using the DRBD device
Let’s create a filesystem on our new DRBD device. I prefer using LVM since that makes it easy to manage the partitions later on. But you may also simply use the /dev/drbd0 device as any block device on its own.

Initialize LVM2:

pvcreate /dev/drbd0
vgcreate redundantstorage /dev/drbd0

We now have a LVM2 volume group called ‘redundantstorage’ on device /dev/drbd0

Create the desired LVM partitions on it like this:

lvcreate -L 1T -n web_files redundantstorage
lvcreate -L 250G -n other_files redundantstorage

The partitions you create are named like the volume group. You can now use ‘/dev/redundantstorage/web_files’ and ‘/dev/redundantstorage/other_files’ like you’d otherwise use ‘/dev/sda3’ etc.

Before we can actually use them, we need to create a file system on top:

mkfs.ext4 /dev/redundantstorage/web_files
mkfs.ext4 /dev/redundantstorage/other_files

Finally, mount the file systems:

mkdir /redundantstorage/web_files
mkdir /redundantstorage/other_files
mount /dev/redundantstorage/web_files /redundantstorage/web_files
mount /dev/redundantstorage/other_files /redundantstorage/other_files

Using the DRBD file systems
Two more steps are needed to set up before we can test our new redundant storage cluster: Heartbeat to manage the cluster and NFS to make use of it. Let’s start with NFS, so Heartbeat will be able to manage that late on as well.

To install NFS server, simply run:

apt-get install nfs-kernel-server

Then setup what folders you want to export using your NFS server.

vim /etc/exports

And enter this configuration:


Pay attention to the the ‘fsid’ parameter. It is really important because it tells the clients that the file system on the primary and secondary are both the same. If you omit this parameter, the clients will ‘hang’ and wait for the old primary to come back online after a fail over happens. Since this is not what we want, we need to tell the clients the other server is simply the same. Fail-over will then happen almost without notice. Most tutorials I read do not tell you about this crucial step.

Make sure you have all this setup on both servers. Since we want Heartbeat to manage our NFS server, we need not to start NFS on boot. To do that, run:

update-rc.d -f nfs-common remove
update-rc.d -f nfs-kernel-server remove

Basic Heartbeat configuration
Install the heartbeat packages is simple:

apt-get install heartbeat

If you’re on CentOS, have a look at the EPEL repository. I’ve successfully setup Heartbeat with those packages as well.

To configure Heartbeat:

vim /etc/ha.d/

Enter this configuration:

autojoin none
auto_failback off
keepalive 2
warntime 5
deadtime 10
initdead 20
bcast eth0
logfile /var/log/heartbeat-log
debugfile /var/log/heartbeat-debug

I set ‘auto_failback’ to off, since I do not want another fail-over when the old primary comes back. If your primary server has better hardware than the secondary one, you may want to set this to ‘on’ instead.

The parameter ‘deadtime’ tells Heartbeat to declare the other node dead after this many seconds. Heartbeat will send a heartbeat every ‘keepalive’ number of seconds.

Protect your heartbeat setup with a password:

echo "auth 3
3 md5 your_secret_password
" > /etc/ha.d/authkeys
chmod 600 /etc/heartbeat/authkeys

You need to select an ip-address that will be your ‘service’-address. Both servers have their own 10.10.0.x ip-address, so choose another one in the same range. I use in this example. Why we need this? Simply because you cannot know to which server you should connect. That’s why we will instruct Heartbeat to manage an extra ip-address and make that alive on the current primary server. When clients connect to this ip-address it will always work.

In the ‘haresources’ file you describe all services Heartbeat manages. In our case, these services are:
– service ip-address
– DRBD disk
– LVM2 service
– Two filesystems
– NFS daemons

Enter them in the order they need to start. When shutting down, Heartbeat will run them in reversed order.

vim /etc/ha.d/haresources

Enter this configuration: \
IPaddr:: \
drbddisk::redundantstorage \
lvm2 \
Filesystem::/dev/redundantstorage/web_files::/redundantstorage/web_files::ext4::nosuid,usrquota,noatime \
Filesystem::/dev/redundantstorage/other_files::/redundantstorage/other_files::ext4::nosuid,usrquota,noatime \
nfs-common \

Use the same Heartbeat configuration on both servers. In the ‘haresources’ file you specify one of the nodes to be the primary. In our case it’s ‘storage-server0’. When this server is or becomes unavailable, Heartbeat will start the services it knows on the other node, ‘storage-server1’ in this case (as specified in the config file).

Wrapping up
DRBD combined with Heartbeat and NFS creates a powerful, redundant storage solution all based on Open Source software. When using the right hardware you will be able to achieve great performance with this setup as well. Think about RAID controllers with SSD-cache and don’t forget the Battery Backup Unit so you can enable the Write Back Cache.

Enjoy building your redundant storage!