You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
DoTheEvo e986b0a93f update 3 days ago
kopia_cli_deploy_win update 2 months ago
kopia_server_deploy_service_win update 2 months ago update 3 days ago



Last update april 2024 kopia v0.17.0



Purpose & Overview


Kopia is a new open source backup utility with basically all modern features.
Cross-platform, deduplication, encryption, compression, multithreaded speed, native cloud storage support, repository replication, snapshots mounting, GUI version, server version,...

Written in golang.
Embedded webGUI for server mode is done in React. KopiaUI comes packaged with electron.

Ways to use Kopia

  • cli - Command line.
    You execute kopia binary passing some commands, it does stuff, done.
    Deployment requires extra work - scripts with configs, scheduling.
  • KopiaUI - GUI version.
    Kopia that comes packaged with electron for GUI to provide the feel of a standalone desktop app.
    Good for simple deployment where average user just wants to backup stuff.
    Benefits over cli is much easier setup, management.
    Drawback is that it runs under one user and only when that user is logged in.
  • Kopia Server - kopia binary runs in a server mode.
    Runs in the background, now with its web server answering at url: localhost:51515
    Web GUI at the url makes the management easy so it can replace KopiaUI.
    But the intended purpose of the server mode is that Kopia instance can now serve as a centralized repository for other machines that run their Kopia instances and select server's url as a repository where to backup.
    Deployment requires extra work similar to cli, but actual use is through web GUI.
  • Kopia in Docker - Kopia Server running as a docker container
    Can fulfill two needs:
    • A centralized Kopia repository where other machines on the network, that also use Kopia, backup their data.
    • Backup docker-host stuff to a cloud or a mounted network storage. Managed through webgui instead of cli.


Some aspects of Kopia

Official Getting Started Guide
Advanced Topics

The above linked documentation is well written and worth a look if planning serious use.

  • Kopia is a single ~35MB binary file.
  • Backups are stored in a repository that needs to be created first, and is always encrypted.
  • Before any action, Kopia needs to be connected to a repo as repos store most of the settings(policies), and commands are executed in their context. Multiple machines can be connected simultaneously.
  • Snapshots are the backups created by kopia, stored in a repository.
  • Policy - settings for repo/backup behaviour, stuff like backups retention, what to ignore, logging, scheduling(server/UI), actions before and after backup,...
  • Policies are stored inside a repo and can apply at various levels and can inherit from each other
    • Global policy, the default that comes predefined during repo creation, can be edited like any other.
    • Per user@machine policy
    • Snapshot level policy, only applying for that one path.
  • Maintenance is automatic.
  • Retention of backups - here's how it works under the hood.
  • Restore from backups is most easily done by mounting a snapshot.
    Web GUI versions have button for it, cli version can do sudo kopia mount all /mnt/temp &
  • Tasks section in gui gets wiped when Kopia closes, info on snapshots run history and duration then has to be find in logs
  • Logs are creted on every execution of kopia binary.
    They rotate by default with max age 30 days, but still can grow hundreds of MB.
  • Compression is good and should be set before backup starts. My go-to is zstd-fastest. If backups feel slow s2-default is less cpu heavy but with worse compression. Useful command: kopia content stats
  • During snapshots Kopia uses local cache, location varies depending on the OS. Default max size is 5GB. Cache gets swept periodically every few minutes.
    Useful commands are kopia cache info and kopia cache clear.
    Increase considerably the max cache size if planning to use cloud storage, as the maintenance could eat into egress cost when kopia redownloads files.
  • ...

Kopia in Linux


A script will be periodically executing cli version of kopia to connect to a repository, execute backup, and disconnect.
Systemd-timers are used to schedule execution of the script. The repository is created on a network share, also mounted on boot using systemd.

Install Kopia

For arch linux, kopia is on AUR yay kopia-bin
For other distros

The initial steps and general use commands

  • repo creation

sudo kopia repo create filesystem --path /mnt/mirror/KOPIA/docker_host_kopia
sudo kopia repo connect filesystem --path /mnt/mirror/KOPIA/docker_host_kopia
sudo kopia repo status

If the path used during creation does not exists, kopia will create it in full.
After creation the repo is connected, so connnect command is just demonstration.

  • the policy info and change

sudo kopia policy list
sudo kopia policy show --global
sudo kopia policy set --global --compression=zstd-fastest --keep-annual=0 --keep-monthly=12 --keep-weekly=0 --keep-daily=14 --keep-hourly=0 --keep-latest=3

  • manual backup run

sudo kopia snapshot create /home/spravca/docker /etc
sudo kopia snapshot list

Since the connection exists with a repo, all that is needed is target that should be backed up.

  • mounting backups

sudo kopia mount all /mnt/tmp & - mounts all snapshots
sudo kopia snapshot list
sudo kopia mount k7e2b0a503edd7604ff61c68655cd5ad7 /mnt/tmp &
sudo umount /mnt/tmp

The backup script

In linux, passing multiple paths separated by space seems to work fine.
So both /home and /etc are set to be backed up.



# v0.2
# initialize repository
#   sudo kopia repo create filesystem --path /mnt/mirror/KOPIA/docker_host_kopia
# for cloud like backblaze
#   sudo kopia repository create b2 --bucket=rakanishu --key-id=001496285081a7e0000000003 --key=K0016L8FAMRp/F+6ckbXIYpP0UgTky0
#   sudo kopia repository connect b2 --bucket=rakanishu --key-id=001496285081a7e0000000003 --key=K0016L8FAMRp/F+6ckbXIYpP0UgTky0
# adjust global policy
#   sudo kopia policy set --global --compression=zstd-fastest --keep-annual=0 --keep-monthly=12 --keep-weekly=0 --keep-daily=14 --keep-hourly=0 --keep-latest=3

BACKUP_THIS='/home /etc'
export KOPIA_PASSWORD='aaa'

kopia repository connect filesystem --path $REPOSITORY_PATH
kopia snapshot create $BACKUP_THIS
kopia repository disconnect

# --------------  ERROR EXIT CODES  --------------
# kopia does not interupts its run with an error if target or repository are missing
# this makes systemd OnSuccess OnFailure not behaving as one might expect
# below are checks for paths, that result in immediate error exit code if they do not exist
# they are at the end because some backup might get done even if another is missing something
# we just want the error exit code

IFS=' ' read -ra paths <<< "$BACKUP_THIS"
for path in "${paths[@]}"; do
  if [ ! -e "$path" ]; then
    echo "ERROR: Backup target '$path' does not exist."
    exit 1

if [ ! -d "$REPOSITORY_PATH" ]; then
  echo "ERROR: Directory '$REPOSITORY_PATH' does not exist."
  exit 1

make the script executable
sudo chmod +x /opt/

Scheduled backups using systemd

Usually cron is used, but systemd provides better logging and control, so better get used to using it.
Heres some discussion on unit files.
ntfy can be used for notifications, more info here

  • sudo micro /etc/systemd/system/kopia-home-etc.service


Description=kopia backup
# OnFailure=ntfy@failure-%p.service
# OnSuccess=ntfy@success-%p.service


# Lower CPU and I/O priority.


  • sudo micro /etc/systemd/system/kopia-home-etc.timer


Description=Run kopia backup

OnCalendar=*-*-* 02:00:00

  • sudo systemctl enable --now kopia-home-etc.timer
  • systemctl status kopia-home-etc.timer
  • journalctl -u kopia-home-etc.timer - see history


To see logs of last Kopia runs done by systemd

  • sudo journalctl -ru kopia-home-etc.service
  • sudo journalctl -xru kopia-home-etc.service


Mounting network storage using systemd

  • files are placed in /etc/systemd/system
  • the name of mount and automount files MUST correspond with the path
    replacing / with a -, but otherwise it must be the mounting path in the name
  • for mounting that does not fail on boot if there are network issues, and mounts the target only on request - enable automount file, not mount file, so:
    sudo systemctl enable mnt-mirror.automount


Description=3TB truenas mirror mount




Description=3TB truenas mirror mount



Kopia in Windows


KopiaUI in Windows

KopiaUI does not really need a guide. It's simple and just works for normal use.
But since we are here...

  • Download latest release
  • Extract it somewhere, lets say C:\Kopia
  • Run it, click through repo creation
  • set global policy
    • recommend setting compression, zstd-fastest
    • set schedule
    • retention rules
  • select what to backup
  • Right click tray icon and set to "Launch at startup"
  • done

It will now start on users login, and executes at set schedule.

While KopiaUI seems like the way to go because of the simple deployment and use, it has a drawback. The scheduled backups works only when user is logged in. Which for many deployments feel like it introduces unnecessary uncertainty, or is not even viable when servers often run with no user logged in.

Kopia Server in Windows

My go-to for windows use because it offers gui for easier managment and the reliability of always running in the background as a service.

This deployment does not make use of the main Kopia Server feature, to be a repository for other machines running Kopiam, just local deployment. Few edits of kopia_server_start.cmd can make it happen though.

  • Download this repo, delete everything except kopia_backup/kopia_server_deploy_service_win folder.
  • Run DEPLOY.cmd as admin, it will:
    • Removes powershell scripts restriction.
    • Creates folder C:\Kopia and copies files there.
    • Uses shawl to create Kopia service.
    • Places kopia.url on the current user's desktop.
  • One should check content of C:\Kopia\kopia_server_start.cmd
    that's where credentials are set, default: admin // aaa
  • Visit in browser localhost:51515
  • Setup new repo through webgui.
  • set global policy
    • recommend setting compression, zstd-fastest
    • set schedule
    • retention rules
  • Select what to backup.

Kopia should now run on boot and be easy to manage through web GUI.
Be it creating backup jobs, mounting old snapshots to restore files, or just looking around if all works as it should.

All relevant files are in C:\Kopia, from binaries, repository.config, to logs.

Before shawl, task scheduler was used.
This matushorvath/Kopia as Windows Service guide helped move beyond that. It contains more info if one would want to actually run as server repository for other machines.
Also use of nssm is popular.

Kopia cli in Windows


At the moment cli is the only way to use VSS snapshots.

All relevant files are in C:\Kopia, from binaries, repository.config, to logs. A scheduled task is imported that executes a powershell script C:\Kopia\kopia_backup_scipt.ps1 at 21:19. The script connects to a set repo and backup set targets.

This approach is bit more hands on than having a gui, but for daily use one can easily get by with the commands: kopia snap list -all and kopia mount all K:
Note that if mount command is not working, try executing it in non admin terminal. Weird windows thing. Or you need to enable/install WebClient service.

  • Download this repo, delete everything except kopia_cli_deploy_win folder.
  • Run DEPLOY.cmd
    • Removes powershell scripts restriction.
    • Creates folder C:\Kopia and kopies there
      kopia.exe, kopia_backup_scipt.ps1.
    • Adds C:\Kopia to the system env variable PATH.
    • imports a scheduled task.
  • Read kopia_backup_scipt.ps1 and follow the instructions there.
    Which should be to just to create repo before running the script.
  • edit the scheduled task to the prefered time, default is daily at 21:19
  • run scheduled task manually
  • check if it worked
    • kopia repo status
    • kopia snap list --all

The script is set to save logs in to C:\Kopia\Kopia_Logs\.

VSS snapshots

Volume Shadow Copy Service freezes the state of the disk in time and makes this snapshot available to use. This is what allows backup of open files that are in use.
Here's some youtube video on VSS.

To make use of this feature edit kopia_backup_scipt.ps1 changing $USE_SHADOW_COPY = $false to $USE_SHADOW_COPY = $true

To check if it's set: kopia policy show --global, should see there: OS-level snapshot support: Volume Shadow Copy: when-available

Can also check log files, any named snapshot-creat in cli folder, and see entries about volume shadow copy. Or also one might execute command vssadmin list shadows to see current VSS snapshots and then execute it again during the backup.

Kopia install using scoop, machine-wide

Just something to have note of, if decided to switch to heavy scoop use.

  • open terminal as admin
  • Set-ExecutionPolicy Bypass
  • iex "& {$(irm} -RunAsAdmin"
  • scoop install sudo --global
  • sudo scoop install kopia --global

Kopia in Docker


Files and directory structure

└── mirror/
└── ~/
    └── docker/
        └── kopia/
            ├── kopia_config/
            ├── kopia_cache/
            ├── kopia_logs/
            ├── some_data/
            ├── kopia_repository/
            ├── kopia_tmp/
            ├── .env
            └── docker-compose.yml
  • /mnt/mirror/... - a mounted network storage share
  • kopia_config/ - repository.config and ui-preferences.json
  • kopia_cache/ - cache
  • kopia_logs/ - logs
  • some_data/ - some data to be backed up
  • kopia_repository/ - repository location
  • kopia_tmp/ - temp used for snapshots
  • .env - a file containing environment variables for docker compose
  • docker-compose.yml - a docker compose file, telling docker how to run the containers


The data to be backed up are mounted in read only mode.
To be able to mount snapshots, extra privileges are required and fuse access.


    image: kopia/kopia:latest
    container_name: kopia
    hostname: kopia
    restart: unless-stopped
    env_file: .env
    privileged: true
      - SYS_ADMIN
      - apparmor:unconfined
      - /dev/fuse:/dev/fuse:rwm
      - "51515:51515"
      - server
      - start
      - --tls-generate-cert
      - --disable-csrf-token-checks
      - --address=
      - --server-username=$USERNAME
      - --server-password=$KOPIA_PASSWORD
        # Mount local folders needed by kopia
        - ./kopia_config:/app/config
        - ./kopia_cache:/app/cache
        - ./kopia_logs:/app/logs
        # Mount local folders to snapshot
        - ./some_data:/data:ro
        # Mount repository location
        - /mnt/mirror/kopia_repository:/repository
        # Mount path for browsing mounted snaphots
        - ./kopia_tmp:/tmp:shared

    external: true




All containers must be on the same network.
Which is named in the .env file.
If one does not exist yet: docker network create caddy_net

Reverse proxy

Caddy is used, details here.

To function as a repository server, Kopia must be setup with https, which is achieved by using --tls-generate-cert flag and removal of --insecure flag.

So now Kopia sits behind Caddy, but caddy needs to be told the traffic is https and to ignore that the certificate is not valid.


kopia.{$MY_DOMAIN} {
  reverse_proxy kopia:51515 {
    transport http {

First run


  • visit
  • create new repository as Local Directory or NAS, set path to /repository, set password

Now this container can do backups of mounted stuff in to other mounted places or cloud, while managed through webgui.

To also make it function as a repository server a user account needs to be added. The users are stored in the repo.

  • exec in to the container
    docker container exec -it kopia /bin/bash
  • add user@machine and set the password
    kopia server user add user1@machine1
  • on another machine test with koppiaUI, on the first run:
    • Pick Kopia Repository Server
    • Server address:
    • Trusted server certificate fingerprint (SHA256)
      can be left empty, or if you put something there, it gives you error where it tells you fingerprints of the server to pick from.
    • In advanced option one can override user@machine with the one set when exec-ed in to the docker container. Or exec again there and add another one.


  • check kopia docker container logs, I like using ctop to see container runtime logs, or the ones mounted in logs directory
  • DNS issue, check nslookup if on the machine is getting correct iP
  • Make sure you use port 443 in server address.

Kopia backup to Cloud

Backblaze B2


Cheapest cloud storage I believe.
It cost $6 annualy to store 100GB. Any download is charged extra, that 100GB would cost $1.

Official Kopia documentation

  • Register.
  • Create a new bucket for kopia repository.
    • note the bucket name
  • Add a new application key with the access enabled to the new bucket.
    After filling the info the site one time shows applicationKey
    • note the keyID
    • note the applicationKey
  • in Kopia add new repository Backblaze b2 fill in the required information: Bucket Name, KeyID and the Key.
  • Set global policy.
    • Recommend setting compression, zstd-fastest.
    • Set schedule.
    • Retention rules.
  • Pick what to backup.
  • Done.

In few minutes one can have reliable encrypted cloud backup, that deduplicates and compresses the data.

Save the repo password, plus all the info used!

Might be worth to check bucket settings, Lifecycle. I think it should be set to Keep only the last version of the file

To restore files go in to Snapshots > Time > Start time > Mount as Local Filesystem.
The snapshot will be mounted as Y:


For cli just follow the official documentation. The example of commands:

  • kopia repository create b2 --bucket=rakanishu --key-id=001496285081a7e0000000003 --key=K0016L8FAMRp/F+6ckbXIYpP0UgTky0 --password aaa
  • kopia repository connect b2 --bucket=rakanishu --key-id=001496285081a7e0000000003 --key=K0016L8FAMRp/F+6ckbXIYpP0UgTky0 --password aaa

The backup script contains example commands, just commented out.