Learn DragonFlyBSD - Storage Engineering: HAMMER2 Production

#Introduction

In episode 18 you ran VMs on top of NVMM. Now we apply all the storage knowledge you've gathered since episode 7 to one final, very tangible goal: designing production storage. This is the moment when HAMMER2 — the filesystem that motivated you to learn this series from the start — really does its job.

Imagine you've been assigned to build a company's storage server: many disks, ever-growing data, daily backup requirements, and demands for data integrity. Episode 19 lays out the blueprint: multi-volume HAMMER2 to use all the disks, dedup and encryption for efficiency and security, scheduled snapshots via cron, routine monitoring, and smooth resizing as capacity grows.

#Use Case: NAS / Storage Server

#Multi-Volume Architecture

The basis of production storage is using all available disk space. With multi-volume HAMMER2, additional disks are formatted and joined without stopping services:

newfs_hammer2 -L data -s 4T /dev/da1s1d
newfs_hammer2 -L data -s 4T /dev/da2s1d
mount_hammer2 /dev/da1s1d@data /srv/nas
hammer2 volume-add /dev/da2s1d /srv/nas

After volume-add, the data filesystem has 8 TB from two disks. When you need more, format a new disk and volume-add again — capacity grows hot.

#A PFS Structure per Service

Inside that large filesystem, split it into per-service PFSes so snapshots and properties can be managed per need:

hammer2 pfs-create /srv/nas/docker
hammer2 pfs-create /srv/nas/backup
hammer2 pfs-create /srv/nas/secure
mount_hammer2 /srv/nas/backup /mnt/backup

The secure PFS can be encrypted, the backup PFS gets the most frequent snapshots, and the docker PFS uses zstd compression for its repetitive images.

#Dedup for Space

Dedup works best on repeated data — VM images, tar backups, ISOs. Enable and run scheduled bulkfree:

hammer2 bulkfree /srv/nas

#Encryption for Sensitive Data

For PFSes storing sensitive data, enable encryption and document the keys:

hammer2 set-encrypted /srv/nas/secure

#Best Practices

#Scheduling Snapshots with Cron

Regular snapshots are the backbone of recovery. Set up a schedule with cron — daily snapshots kept for a few days, weekly snapshots kept longer:

# snapshot harian jam 2 pagi, simpan 7 hari
0 2 * * * hammer2 snapshot /srv/nas/docker
0 2 * * * hammer2 snapshot /srv/nas/backup

Remove snapshots past their target age with scheduled hammer2 pfs-destroy, or manage them manually at specific points in time. The key to a good schedule is regularity, not variety.

#Monitoring with hammer2 status

hammer2 status becomes your health dashboard:

hammer2 status /srv/nas
df -h /srv/nas

Integrate it into a monitoring system (for example cron + mail, or output to an external tool). When usage approaches the limit, that's the moment to add a volume or clean up old snapshots.

#growfs: Resizing the Filesystem (6.2+)

When a volume is enlarged — for example a VM's virtual disk is expanded — HAMMER2 can extend the filesystem without unmounting. Since 6.2, growfs handles resizing automatically:

growfs /srv/nas

After the physical disk grows, growfs adjusts the volume size. Check the size after growfs with hammer2 status — the new capacity is immediately available for the next volumes and snapshots.

#Off-Site Backup Strategy

A storage server without off-site backup is a time bomb. Combine:

• Local snapshots — fast recovery from manual mistakes.
• rsync or hammer2 send to another host — protection from disk loss.
• Tar archives to other media — portability for major disasters.

Apply the 3-2-1 rule from episode 11: three copies, two media types, one off-site.

#Closing

In this episode 19 you designed production storage with HAMMER2: building a multi-volume NAS with hammer2 volume-add, arranging per-service PFSes with different properties, enabling dedup and encryption, scheduling snapshots with cron, monitoring with hammer2 status, and expanding the filesystem with growfs (6.2+).

Key takeaways:

• Multi-volume HAMMER2 combines many disks hot: hammer2 volume-add /dev/da2s1d /srv/nas.
• Split storage into per-service PFSes so snapshots and properties can be managed per need.
• Schedule snapshots with cron and test restores periodically.
• hammer2 status and df -h are your daily monitoring dashboards.
• growfs expands the filesystem without unmounting; combine local snapshots + off-site rsync/send.

In the next episode, episode 20, we play with source: build system & kernel customization. You'll build the system from source with make buildworld and make buildkernel, install with make installkernel, understand binary updates with dragonfly-update, and write a custom kernel config in /usr/src/sys.