Skip to content

SnapRAID-BTRFS

Background

This guide builds on the Perfect Media Server setup by using BTRFS for data drives and taking advantage of snapraid-btrfs to manage SnapRAID operations using read-only BTRFS snapshots where possible. One of the main limitations of SnapRAID is that there is a dependence on live data being continuously accessible and unchanging not only for complete parity sync purposes, but also for complete recovery in the event that a drive needs to be rebuilt from parity.

Using snapraid-btrfs, there is no requirement to stop any services or ensure that the live filesystem is free of any new files or changes to existing files.

From the snapraid-btrfs repo:

Why use snapraid-btrfs?

A: A major disadvantage of SnapRAID is that the parity files are not updated in realtime. This not only means that new files are not protected until after running snapraid sync, but also creates a form of "write hole" where if files are modified or deleted, some protection of other files which share the same parity block(s) is lost until another sync is completed, since if other files need to be restored using the snapraid fix command, the deleted or modified files will not be available, just as if the disk had failed, or developed a bad sector. This problem can be mitigated by adding additional parities, since SnapRAID permits up to six, or worked around by temporarily moving files into a directory that is excluded in your SnapRAID config file, then completing a sync to remove them from the parity before deleting them. However, this problem is a textbook use case for btrfs snapshots.

By using read-only snapshots when we do a snapraid sync, we ensure that if we modify or delete files during or after the sync, we can always restore the array to the state it was in at the time the read-only snapshots were created, so long as the snapshots are not deleted until another sync is completed with new snapshots. This use case for btrfs snapshots is similar to using btrfs send/receive to back up a live filesystem, where the use of read-only snapshots guarantees the consistency of the result, while using dd would require that the entire filesystem be mounted read-only to prevent corruption caused by writes to the live filesystem during the backup.

Another benefit of using BTRFS for the data drives is it allows for online drive replacements (i.e. upgrades to larger drives) using btrfs-replace. This means that the MergerFS pool can remain mounted with all data accessible for reads and writes while underyling data drives are being replaced. This guide includes instructions for replacing BTRFS data drives in the array.

Prerequisites

Install Ubuntu 20.04 as per the Manual Install on Bare Metal guide at Perfect Media Server, and follow the steps until the end of the Brand new drives section, resulting in a bare filesystem on each drive.

Drive setup

Formatting the drives

Install BTRFS tools:

apt install btrfs-progs

Format each data drive using BTRFS.

Note

Note that for convenience this guide uses filesystem labels for mountpoints in /etc/fstab. It is user preference to use /dev/disk/by-id, /dev/disk/by-uuid , or other method.

mkfs.btrfs -L mergerfsdisk1 /dev/sdX1

Parity drive(s)

Note

Note that although there's no harm with using BTRFS for the SnapRAID parity, there isn't any benefit from doing so to use snapraid-btrfs. It is recommended to use ext4 for the SnapRAID parity.

Format parity drive(s) using mkfs.ext4 and mount as per the Perfect Media Server installation guide.

Data subvolumes

To use BTRFS snapshots as this guide suggests, the data itself will reside in a BTRFS /data subvolume that needs to be created on each drive. Create mountpoints for the BTRFS root filesystems:

mkdir -p /mnt/btrfs-roots/mergerfsdisk{1,2,3,4}

Add entries in /etc/fstab for the root filesystems:

### /etc/fstab BTRFS root filesystems
LABEL=mergerfsdisk1 /mnt/btrfs-roots/mergerfsdisk1 btrfs defaults 0 0
LABEL=mergerfsdisk2 /mnt/btrfs-roots/mergerfsdisk2 btrfs defaults 0 0
...

Mount each disk and create BTRFS subvolumes on each root filesystem:

mount /mnt/btrfs-roots/mergerfsdisk1
btrfs subvolume create /mnt/btrfs-roots/mergerfsdisk1/data
mount /mnt/btrfs-roots/mergerfsdisk2
btrfs subvolume create /mnt/btrfs-roots/mergerfsdisk2/data
...

Create mountpoints for the array as described at Perfect Media Server. For example:

mkdir /mnt/disk{1,2,3,4}

Add entries in /etc/fstab for data subvolumes:

### /etc/fstab BTRFS data subvolumes
LABEL=mergerfsdisk1 /mnt/disk1 btrfs subvol=/data 0 0
LABEL=mergerfsdisk2 /mnt/disk2 btrfs subvol=/data 0 0
...

Mount the data drives:

mount /mnt/disk1
mount /mnt/disk2
...

Content subvolumes

The SnapRAID .content files do not need to be snapshotted and it is recommended that any .content files stored on the array be in a separate BTRFS subvolumes.

Create subvolumes and mountpoints for .content files:

btrfs subvolume create /mnt/btrfs-roots/mergerfsdisk1/content
btrfs subvolume create /mnt/btrfs-roots/mergerfsdisk2/content
mkdir -p /mnt/snapraid-content/disk{1,2}

Add entries in /etc/fstab for content subvolumes:

### /etc/fstab BTRFS content subvolumes
LABEL=mergerfsdisk1 /mnt/snapraid-content/disk1 btrfs subvol=/content 0 0
LABEL=mergerfsdisk2 /mnt/snapraid-content/disk2 btrfs subvol=/content 0 0
...

BTRFS root filesystem unmount

Once the data and content subvolumes are created and mounted, the BTRFS root filesystem can be unmounted.

umount /mnt/btrfs-roots/mergerfsdisk1
umount /mnt/btrfs-roots/mergerfsdisk2
...

The entries for the BTRFS root filesystems in /etc/fstab can also be commented out.

MergerFS

The fstab steps for the MergerFS pool are unchanged from the Perfect Media Server installation guide. Create a /mnt/storage mountpoint and follow fstab entries section of the guide.

SnapRAID setup

Install SnapRAID as per the Perfect Media Server installation guide. To configure SnapRAID, ensure it points to the correct mount points that were created earlier. Using the example at Perfect Media Server as a basis, and updating for the mountpoints above results in the following:

# SnapRAID configuration file

# Parity location(s)
1-parity /mnt/parity1/snapraid.parity
2-parity /mnt/parity2/snapraid.parity

# Content file location(s)
content /var/snapraid.content
content /mnt/snapraid-content/disk1/snapraid.content
content /mnt/snapraid-content/disk2/snapraid.content

# Data disks
data d1 /mnt/disk1
data d2 /mnt/disk2
data d3 /mnt/disk3
data d4 /mnt/disk4

# Excludes hidden files and directories
exclude *.unrecoverable
exclude /tmp/
exclude /lost+found/
exclude downloads/
exclude appdata/
exclude *.!sync
exclude /.snapshots/

At this point, the system is set up to run native SnapRAID as described in the Perfect Media Server installation guide, the only difference being that the data is stored on BTRFS subvolumes.

The remainder of this guide will discuss how to leverage the BTRFS subvolumes with snapshots, and using those for SnapRAID parity.

Snapper setup

Snapper installation and configuration template

In order to create and work with BTRFS snapshots, snapraid-btrfs uses Snapper. Install it as follows:

apt install snapper

Snapper requires that configuration profiles are created for each subvolume that requires snapshots. It has the ability to take new snapshots and cleanup old ones on a regular basis using timeline policies.

Note

For the purposes of this guide, the timeline-based snapshots are not required for snapraid-btrfs. snapraid-btrfs will create its own snapshots in conjunction with SnapRAID operations.

The default Snapper configuration template will be used as a basis for a minimal template for MergerFS data drives.

cd /etc/snapper/config-templates
cp default mergerfsdisk

To disable timeline-based snapshots, edit the /etc/snapper/config-templates/mergerfsdisk template as follows:

...
# create hourly snapshots
TIMELINE_CREATE="no"
...

Additional config options can be found at the snapper-configs man page.

Snapper profiles for data subvolumes

Create Snapper profiles for each data subvolume created earlier using the mergerfsdisk template.

snapper -c mergerfsdisk1 create-config -t mergerfsdisk /mnt/disk1
snapper -c mergerfsdisk2 create-config -t mergerfsdisk /mnt/disk2
...

The resultant config files can be found at /etc/snapper/configs and the subvolumes that they relate to can be verified by running the following:

snapper list-configs

snapraid-btrfs setup

Install snapraid-btrfs by cloning the Git repo and copying the snapraid-btrfs script to your system.

git clone https://github.com/automorphism88/snapraid-btrfs.git
cd snapraid-btrfs
cp snapraid-btrfs /usr/local/bin
chmod +x /usr/local/bin/snapraid-btrfs

Verify that snapraid-btrfs is successfully able to see Snapper configs for each data subvolume by running the following:

snapraid-btrfs ls

At this point, any snapraid <command> can be run as as snapraid-btrfs <command>. Depending on the command, snapraid-btrfs will either take a snapshot, use an existing latest snapshot, or use the live filesystem before passing that command on to SnapRAID for processing.

Automatic parity calculation - snapraid-btrfs-runner

To automate daily parity sync and scrub operations using snapraid-btrfs, this guide uses snapraid-btrfs-runner based on the upstream snapraid-runner tool that is commonly used for SnapRAID automation.

snapraid-btrfs-runner conducts the same basic tasks as snapraid-runner, including diff, sync, and scrub operations.

Install snapraid-btrfs-runner by cloning the repository:

git clone https://github.com/fmoledina/snapraid-btrfs-runner.git /opt/snapraid-btrfs-runner

Create a configuration file based on the example provided at /opt/snapraid-btrfs-runner/snapraid-btrfs-runner.conf.example. All of the snapraid-runner options are present, with additional options available for snapraid-btrfs and Snapper.

Required config parameters are as follows:

  • snapraid-btrfs.executable location
  • snapper.executable location
  • snapraid.executable location
  • snapraid.config location

Other config parameters of interest are as follows:

  • snapraid-btrfs.cleanup: Upon a successful run of snapraid-btrfs-runner, any interim snapshots created during the process will be removed, leaving only the snapraid-btrfs=synced snapshot. Defaults to true.
  • Other options specified at the Perfect Media Server installation guide.

Scheduling can be set either via cron or SystemD timers. This guide provides a basic SystemD timer as follows.

Contents of /etc/systemd/system/snapraid-btrfs-runner.service:

[Unit]
Description=Run snapraid-btrfs-runner every night

[Service]
Type=oneshot
ExecStart=/usr/bin/python3 /opt/snapraid-btrfs-runner/snapraid-btrfs-runner.py -c /opt/snapraid-btrfs-runner/snapraid-btrfs-runner.conf

Contents of /etc/systemd/system/snapraid-btrfs-runner.timer:

[Unit]
Description=Run snapraid-btrfs-runner every night

[Timer]
OnCalendar=*-*-* 03:00:00
RandomizedDeleySec=30m

[Install]
WantedBy=timers.target

Enable using:

systemctl enable snapraid-btrfs-runner.timer
systemctl start snapraid-btrfs-runner.timer

Online Data Drive Replacements

BTRFS allows for online replacements of data drives using btrfs-replace. This allows for replacing data drives with larger ones without ever having to unmount the MergerFS array. In fact, MergerFS is ignorant to the operation since from its perspective, the member mount point always remains accessible while BTRFS conducts the drive replacement on the backend. The target drive needs to be the same size or larger than the source drive.

Use the following steps to replace a BTRFS drive in the above array, e.g. /mnt/disk1.

Warning

Verify that the correct drive is being replaced. Check both source and target drives using serial numbers in /dev/disk/by-id or similar unique identifiers prior to starting the replacement.

Check the current BTRFS filesystem state:

# btrfs filesystem show /mnt/disk1    
Label: 'mergerfsdisk1'  uuid: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
        Total devices 1 FS bytes used 10.14TiB
        devid    1 size 10.91TiB used 10.17TiB path /dev/sds1

Note

BTRFS filesystem-related commands can be carried out at any mountpoint that the filesystem is mounted on. This includes the root mountpoint or any subvolumes that are mounted. In this example, commands are executed at the /mnt/disk1 subvolume mountpoint created above as it's already conveniently mounted.

On the new drive, create a filesystem table and primary partition as described at Perfect Media Server. Assuming the replacement device is /dev/sdy1, start the BTRFS replacement with the following command:

btrfs replace start 1 /dev/sdy1 /mnt/disk1

where 1 refers to the devid from the above filesystem output. This command will produce no output, and BTRFS will have started the replacement in the background.

To check the status of the replacement:

btrfs replace status -1 /mnt/disk1

which provides the current status and exits. To continuously monitor the replacement status, run the btrfs replace status command with the -1 omitted.

Details of the filesystem during the drive replacement will appear as follows:

# btrfs filesystem show /mnt/disk1
Label: 'mergerfsdisk1'  uuid: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
        Total devices 2 FS bytes used 10.14TiB
        devid    0 size 10.91TiB used 10.17TiB path /dev/sdy1
        devid    1 size 10.91TiB used 10.17TiB path /dev/sds1

Once the drive replacement is successfully completed, the above btrfs replace status command would result in the following output:

# btrfs replace status /mnt/disk1
Started on 26.Jan 20:14:52, finished on 27.Jan 12:43:39, 0 write errs, 0 uncorr. read errs

Details of the filesystem after the drive replacement will appear as follows:

# btrfs filesystem show /mnt/disk1
Label: 'mergerfsdisk1'  uuid: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
        Total devices 1 FS bytes used 10.14TiB
        devid    1 size 10.91TiB used 10.17TiB path /dev/sdy1

If the target drive is larger than the source drive, the BTRFS filesystem will need to be resized to take advantage of the full space:

$ btrfs filesystem resize 1:max /mnt/disk1
Resize '/mnt/disk1' of '1:max'

where 1 refers to the devid from above (which is now the new device) and max indicates that the filesystem should use all the available space on the drive.

The final details of the filesystem will now appear as follows:

# btrfs filesystem show /mnt/disk1
Label: 'mergerfsdisk1'  uuid: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
        Total devices 1 FS bytes used 10.14TiB
        devid    1 size 12.73TiB used 10.17TiB path /dev/sdy1

At this point, the replacement operation is complete and the MergerFS pool should be able to see the additional space.