# Setting Up Storage
ArchiveBox supports a wide range of local and remote filesystems using `rclone` and/or Docker storage plugins. The examples below use [Docker Compose bind mounts](https://docs.docker.com/storage/bind-mounts/) to demonstrate the concepts; adapt the host paths, ownership, and provider settings to your environment.
Example [`docker-compose.yml`](https://github.com/ArchiveBox/ArchiveBox/blob/dev/docker-compose.yml) storage setup:
```yaml
services:
archivebox:
# ...other service settings...
volumes:
# your index db, config, logs, etc. should be stored on a local SSD (usually <10Gb)
- ./data:/data
# but bulk archive/ content can be located on an HDD or remote filesystem
- /mnt/archivebox-archive:/data/archive
```
Related Docs
---
## Supported Local Filesystems
### `EXT4` (default on Linux), `APFS` (default on macOS)
> [!TIP]
> These default filesystems are fully supported by ArchiveBox on Linux and macOS (w/wo Docker).
### `ZFS` (recommended for experienced Linux/BSD operators) ⭐️
> [!TIP]
> *ZFS is a good choice when you already operate OpenZFS and want checksumming, compression, snapshots, replication, and optional encryption or disk redundancy.*
> On Ubuntu, follow the official [OpenZFS installation guide](https://openzfs.github.io/openzfs-docs/Getting%20Started/Ubuntu/index.html). macOS and BSD installation and property support differ, so use the guide for your OS.
- https://openzfs.github.io/openzfs-docs/
- https://openzfs.github.io/openzfs-docs/man/v2.2/8/zpool-create.8.html
- https://openzfs.github.io/openzfs-docs/man/v2.2/8/zfs-create.8.html
- https://docs.docker.com/storage/storagedriver/zfs-driver/
- https://www.ixsystems.com/blog/fast-dedup-is-a-valentines-gift-to-the-openzfs-and-truenas-communities/
> [!CAUTION]
> Creating a pool erases the selected disks. The two-disk example below creates a mirror; replace both `/dev/disk/by-id/...` placeholders with the persistent IDs of empty disks you intend to erase.
```bash
# Create a mirrored pool without forcing ZFS's safety checks.
sudo zpool create \
-O mountpoint=none \
-O compression=lz4 \
-O dnodesize=auto \
-O atime=off \
-O xattr=sa \
-O acltype=posixacl \
-O aclinherit=passthrough \
archivebox mirror \
/dev/disk/by-id/DISK_ONE \
/dev/disk/by-id/DISK_TWO
# Create the unencrypted ArchiveBox data dataset.
sudo zfs create \
-o mountpoint=/mnt/archivebox/data \
archivebox/data
```
To encrypt a new dataset, use this command **instead of** the unencrypted `zfs create` command above. ZFS encryption must be selected when the dataset is created.
```bash
sudo zfs create \
-o mountpoint=/mnt/archivebox/data \
-o encryption=on \
-o keyformat=passphrase \
-o keylocation=prompt \
archivebox/data
```
### `NTFS`, `HFS+`, `BTRFS`
> [!WARNING]
> These filesystems are likely supported, but are not officially tested.
### `EXT2`, `EXT3`, `FAT32`, `exFAT`
> [!CAUTION]
> Not recommended. Cannot store files >4GB or more than 31k ~ 65k Snapshot entries due to directory entry limits.
---
## Supported Remote Filesystems
ArchiveBox supports many common types of remote filesystems using Rclone, FUSE, Docker storage providers, and Docker volume plugins.
The `data/archive/` subfolder contains the bulk archived content, and it supports being stored on a slower remote server (SMB/NFS/SFTP/etc.) or object store (S3/B2/R2/etc.). For data integrity and performance reasons, the rest of the `data/` directory (`data/ArchiveBox.conf`, `data/logs`, etc.) must be stored locally while ArchiveBox is running.
> [!IMPORTANT]
> `data/index.sqlite3` is your main archive DB, *it must be on a fast, reliable, local filesystem* which supports [FSYNC](https://stackoverflow.com/questions/40849596/git-clone-fsync-input-output-error-in-linux#:~:text=Some%20filesystems%20%2D%20especially%20remote%20filesystems%20like%20NFS%2C%20sshfs%2C&text=do%20not%20support%20fsync()%20but%20git%20has%20no%20flag%20to%20disable%20these%20calls) (SSD/NVMe recommended for best experience).
> [!TIP]
> If you use a remote filesystem, you should switch ArchiveBox's search backend from [`ripgrep`](https://github.com/ArchiveBox/ArchiveBox/wiki/Setting-up-Search#ripgrep) to [`sonic`](https://github.com/ArchiveBox/ArchiveBox/wiki/Setting-up-Search#sonic) (or [`FTS5`](https://github.com/ArchiveBox/ArchiveBox/wiki/Setting-up-Search#fts5)).
> (`ripgrep` scans over every byte in the archive to do each search, which is **slow and potentially costly** on remote cloud storage)
### `NFS` (Docker Driver)
`docker-compose.yml`:
```yaml
services:
archivebox:
volumes:
- ./data:/data
- archivebox-archive:/data/archive
volumes:
archivebox-archive:
driver: local
driver_opts:
type: "nfs"
o: "addr=some-remote-server.example.com,rw,nfsvers=4"
device: ":/archivebox-archive"
```
### `SMB` / `Ceph` (Docker CIFS Driver)
`docker-compose.yml`:
```yaml
services:
archivebox:
volumes:
- ./data:/data
- archivebox-archive:/data/archive
volumes:
archivebox-archive:
driver: local
driver_opts:
type: cifs
device: "//some-remote-server.example.com/archivebox-archive"
o: "username=XXX,password=YYY,uid=911,gid=911"
```

### Amazon S3 / Backblaze B2 / Google Drive / etc. (RClone)
ArchiveBox stores snapshot content under `data/archive/users//snapshots////` and keeps backwards-compatible `data/archive/` symlinks. Object-storage mounts must enable Rclone's VFS symlink translation so both parts of this layout work.
Install the `rclone` binary through `abxpkg`:
```bash
uv tool install abxpkg
abxpkg install rclone
abxpkg run rclone version
```
Then install the FUSE 3 system integration supplied by your OS. For example, on Ubuntu:
```bash
sudo apt-get install fuse3
grep -qxF user_allow_other /etc/fuse.conf ||
printf '%s\n' user_allow_other | sudo tee -a /etc/fuse.conf
```
Then define your remote storage config `~/.config/rclone/rclone.conf`:
> [!TIP]
> You can also create `rclone.conf` using the Rclone Web GUI: `abxpkg run rclone rcd --rc-web-gui`
```ini
# Example rclone.conf using Amazon S3 for storage:
[archivebox-s3]
type = s3
provider = AWS
access_key_id = XXX
secret_access_key = YYY
region = us-east-1
```
#### Rclone Config Examples
- [SMB](https://rclone.org/smb/) / [Ceph](https://rclone.org/s3/#ceph) / [SFTP](https://rclone.org/sftp/) / [FTP](https://rclone.org/ftp/) / [WebDAV (e.g. Nextcloud)](https://rclone.org/webdav/)
- [Google Drive](https://rclone.org/drive/) / [Dropbox](https://rclone.org/dropbox/) / [OneDrive](https://rclone.org/onedrive/)
- [Amazon S3](https://rclone.org/s3/#configuration) / [Backblaze B2](https://rclone.org/b2/) / [Cloudflare R2](https://rclone.org/s3/#cloudflare-r2) / [DigitalOcean Spaces](https://rclone.org/s3/#digitalocean-spaces)
- [Google Cloud Storage](https://rclone.org/s3/#google-cloud-storage) / [Azure Blob](https://rclone.org/azureblob/) / [Azure Files](https://rclone.org/azurefiles/)
- [Storj](https://rclone.org/s3/#storj) / [Sia](https://rclone.org/sia/) / [Archive.org Storage](https://rclone.org/internetarchive/)
- And many more...
- https://rclone.org/s3/
- https://rclone.org/overview/
*Bonus:*
- Set up gzip compression: https://rclone.org/compress/
- Set up file encryption: https://rclone.org/crypt/
- Set up hashing engine: https://rclone.org/hasher/
#### Option A: Running Rclone on a bare-metal host
1. *If Needed:* Transfer any existing local archive data to the remote volume first
> [!CAUTION]
> Stop ArchiveBox before migrating its archive directory. `rclone sync` makes the remote destination match the local source and can delete files already present at the destination. Run it with `--dry-run` first, make a separate backup, and do not move the local copy until `rclone check` succeeds.
```bash
abxpkg run rclone sync \
--dry-run \
--links \
--fast-list \
--transfers 20 \
--progress \
/opt/archivebox/data/archive/ \
archivebox-s3:data/archive/
# Remove --dry-run only after reviewing the proposed changes, then verify them.
abxpkg run rclone sync \
--links \
--fast-list \
--transfers 20 \
--progress \
/opt/archivebox/data/archive/ \
archivebox-s3:data/archive/
abxpkg run rclone check --links /opt/archivebox/data/archive/ archivebox-s3:data/archive/
mv /opt/archivebox/data/archive /opt/archivebox/data/archive.localbackup
mkdir -p /opt/archivebox/data/archive
```
2. **Mount the remote storage volume as FUSE filesystem**
Run the mount as the numeric user that owns the local ArchiveBox collection. The command stays in the foreground so a service manager can supervise it.
```bash
abxpkg run rclone mount \
archivebox-s3:data/archive/ \
/opt/archivebox/data/archive/ \
--allow-other \
--vfs-cache-mode=full \
--vfs-links \
--transfers=16 \
--checkers=4
```
See [Rclone's `rclone mount` documentation](https://rclone.org/commands/rclone_mount/) for service-manager and cache-size configuration.
> [!TIP]
> You can use an existing Rclone FUSE mount as a normal Docker bind mount. A separate storage plugin is usually unnecessary when `user_allow_other` and `--allow-other` are configured correctly.
```bash
docker run --rm \
-v "$PWD:/data" \
-v /opt/archivebox/data/archive:/data/archive \
archivebox/archivebox:dev status
```
`docker-compose.yml`:
```yaml
services:
archivebox:
# ...other service settings...
volumes:
- ./data:/data
- /opt/archivebox/data/archive:/data/archive
```
#### Option B: Running Rclone with the Docker storage plugin
*This Linux Docker Engine option is only needed if you cannot use Option A for compatibility or performance reasons, or if you prefer defining your remote storage in `docker-compose.yml`.*
See here for full instructions: [Rclone Documentation: Docker Plugin](https://rclone.org/docker/)
1. First, install the [Rclone Docker Volume Plugin](https://rclone.org/docker/#installing-as-managed-plugin) for your CPU architecture (e.g. `amd64` or `arm64`):
```bash
sudo mkdir -p \
/var/lib/docker-plugins/rclone/config \
/var/lib/docker-plugins/rclone/cache
sudo install -m 600 \
~/.config/rclone/rclone.conf \
/var/lib/docker-plugins/rclone/config/rclone.conf
# Replace amd64 with arm64 on ARM hosts.
docker plugin install rclone/docker-volume-rclone:amd64 --grant-all-permissions --alias rclone
```
2. Then, [create a volume using the Docker CLI](https://rclone.org/docker/#creating-volumes-via-cli) or [define one using Docker Compose / Swarm](https://rclone.org/docker/#using-with-swarm-or-compose):
`docker-compose.yml`:
```yaml
services:
archivebox:
volumes:
- ./data:/data
- archivebox-s3:/data/archive
volumes:
archivebox-s3:
driver: rclone
driver_opts:
remote: 'archivebox-s3:data/archive'
allow_other: 'true'
vfs_cache_mode: full
vfs_links: 'true'
# Match these to the numeric owner of ./data; 911:911 is the image default.
uid: 911
gid: 911
transfers: 16
checkers: 4
```
To start the container and verify the filesystem is accessible within it:
```bash
docker compose run --rm archivebox \
/bin/bash -c 'touch /data/archive/.write_test && rm /data/archive/.write_test'
```
---
### More Docker Storage Plugins
- [IPFS](https://github.com/djdv/go-filesystem-utils/pull/40) / [Peergos](https://github.com/peergos/peergos) / [GlusterFS](https://github.com/calavera/docker-volume-glusterfs)
- [DigitalOcean Block Storage Volumes](https://github.com/djmaze/dobs-volume-plugin) / [Linode Block Storage Volumes](https://github.com/linode/docker-volume-linode)
- [More volume plugins...](https://docs.docker.com/engine/extend/legacy_plugins/#volume-plugins)