docs/guide/podman.md
Dozzle supports Podman through its Docker-compatible socket interface. Two known differences from Docker that affect setup: memory stats are often missing in rootless/Quadlet deployments (cgroup delegation), and Podman doesn't generate an engine-id. This guide covers standalone mode (local monitoring) and agent mode (remote monitoring via a central Dozzle server).
| Mode | Use Case | Setup Complexity |
|---|---|---|
| Standalone | Single host log viewing | Simple |
| Agent | Multi-host centralized monitoring | Moderate |
Podman offers several launch approaches:
| Method | Auto-start | Memory Stats | Healthchecks | Best For |
|---|---|---|---|---|
| CLI | Manual | ✓ | ✓ | Development |
podman-compose | ✗ | ✓ | ✗ | Testing |
| Quadlet (systemd) | ✓ | ✗* | ✓ | Production |
*Memory stats are typically unavailable in rootless mode unless cgroup v2 memory delegation is enabled. See the FAQ at the bottom of this page.
Run Dozzle as a standalone service to monitor local Podman containers.
For system-wide Podman daemon:
# Enable and start the Podman socket
sudo systemctl enable podman.socket
sudo systemctl start podman.socket
# Dozzle can connect via the Docker socket
podman run -v /run/podman/podman.sock:/var/run/docker.sock:ro \
-p 3000:8080 \
ghcr.io/amir20/dozzle:latest
Rootless Podman isolates containers to a user namespace:
# Start user-level socket (runs automatically with user session)
systemctl --user enable podman.socket
systemctl --user start podman.socket
# For a user named 'appuser', Dozzle can connect via:
podman run -v /run/user/$(id -u appuser)/podman/podman.sock:/var/run/docker.sock:ro \
-p 3000:8080 \
ghcr.io/amir20/dozzle:latest
Important: A Dozzle bound to one user's rootless socket only sees that user's containers. Other users' rootless containers live in separate namespaces and won't appear.
Quadlet enables systemd-native container management. Create a .container file at ~/.config/containers/systemd/dozzle.container:
[Unit]
Description=Dozzle Log Viewer
After=network-online.target
Wants=network-online.target
[Container]
Image=ghcr.io/amir20/dozzle:latest
PublishPort=3000:8080
Volume=/run/user/%U/podman/podman.sock:/var/run/docker.sock:ro
HealthCmd=/dozzle healthcheck
HealthInterval=5s
HealthTimeout=10s
HealthRetries=5
HealthStartPeriod=15s
[Service]
Restart=on-failure
RestartSec=10
[Install]
WantedBy=default.target
Enable and start:
systemctl --user daemon-reload
systemctl --user enable --now dozzle.service
For multi-user systems, drop the same file into each user's ~/.config/containers/systemd/ and pick a distinct host port per user (e.g. PublishPort=3001:8080). Each instance only sees that user's rootless containers.
[!NOTE] Quadlet generates a systemd timer for healthchecks.
podman-composedoes not, so healthchecks won't run on a schedule there; trigger them manually withpodman healthcheck run NAMEif needed.
Run Dozzle as an agent on remote Podman hosts for centralized monitoring via a main Dozzle server. Agents communicate with the main server via gRPC.
Run Dozzle in agent mode on remote Podman hosts:
# Rootful agent
podman run -d \
--name dozzle-agent \
-v /run/podman/podman.sock:/var/run/docker.sock:ro \
-p 7007:7007 \
ghcr.io/amir20/dozzle:latest agent
# Rootless agent (for user 'appuser')
sudo -u appuser podman run -d \
--name dozzle-agent \
-v /run/user/$(id -u appuser)/podman/podman.sock:/var/run/docker.sock:ro \
-p 7007:7007 \
ghcr.io/amir20/dozzle:latest agent
Create a .container file for the agent:
# dozzle-agent.container
[Unit]
Description=Dozzle Agent
After=network-online.target
Wants=network-online.target
[Container]
Image=ghcr.io/amir20/dozzle:latest
PublishPort=7007:7007
Volume=/run/user/%U/podman/podman.sock:/var/run/docker.sock:ro
Exec=agent
HealthCmd=/dozzle healthcheck
HealthInterval=5s
HealthTimeout=10s
HealthRetries=5
HealthStartPeriod=15s
[Service]
Restart=on-failure
RestartSec=10
[Install]
WantedBy=default.target
[!NOTE] The Dozzle image's entrypoint is
/dozzle, soagentgoes inExec=(the command), notEntrypoint=.
Enable and start:
systemctl --user daemon-reload
systemctl --user enable dozzle-agent.service
systemctl --user start dozzle-agent.service
Configure the main Dozzle server to connect to agents on remote Podman hosts.
Run the main Dozzle server with agent endpoints:
podman run -d \
--name dozzle \
-p 3000:8080 \
ghcr.io/amir20/dozzle:latest \
--agent "host1.example.com:7007" \
--agent "host2.example.com:7007"
Or with environment variables:
podman run -d \
--name dozzle \
-e DOZZLE_REMOTE_AGENT="host1.example.com:7007,host2.example.com:7007" \
-p 3000:8080 \
ghcr.io/amir20/dozzle:latest
# dozzle-server.container
[Unit]
Description=Dozzle Server with Remote Agents
After=network-online.target
Wants=network-online.target
[Container]
Image=ghcr.io/amir20/dozzle:latest
PublishPort=3000:8080
Environment=DOZZLE_REMOTE_AGENT=host1.example.com:7007,host2.example.com:7007
HealthCmd=/dozzle healthcheck
HealthInterval=5s
HealthTimeout=10s
HealthRetries=5
HealthStartPeriod=15s
[Service]
Restart=on-failure
RestartSec=10
[Install]
WantedBy=default.target
[!NOTE]
WantedBy=multi-user.targetonly applies to system units. Forsystemctl --userunits, usedefault.target.
Podman doesn't create an engine-id like Docker. Create one to avoid "host not found" errors:
# Create directory if needed
sudo mkdir -p /var/lib/docker
# Generate UUID
sudo sh -c 'uuidgen > /var/lib/docker/engine-id'
# Verify
cat /var/lib/docker/engine-id
- name: Create /var/lib/docker
ansible.builtin.file:
path: /var/lib/docker
state: directory
mode: "755"
- name: Create engine-id and derive UUID from hostname
ansible.builtin.lineinfile:
path: /var/lib/docker/engine-id
line: "{{ hostname | to_uuid }}"
create: true
mode: "0644"
insertafter: "EOF"
[!WARNING] Clean up existing Dozzle deployments (stop container, remove volumes) before recreating with the engine-id in place.
Memory stats are usually missing in rootless deployments because the memory cgroup controller isn't delegated to the user slice by default. Check what's delegated:
cat /sys/fs/cgroup/user.slice/user-$(id -u).slice/cgroup.controllers
If memory is not in the output, enable delegation via a drop-in:
sudo mkdir -p /etc/systemd/system/[email protected]
sudo tee /etc/systemd/system/[email protected]/delegate.conf <<'EOF'
[Service]
Delegate=cpu cpuset io memory pids
EOF
sudo systemctl daemon-reload
Then log out and back in (or reboot) for the user slice to pick up the new delegation. See the Podman rootless tutorial for details.
podman-compose issue: Healthchecks are reported as unhealthy even though manual runs pass. This is a Podman behavior where healthchecks aren't automatically evaluated without a systemd timer (Quadlet generates one automatically).
Workaround with podman-compose:
# Manual healthcheck run
podman healthcheck run <container_id>
Quadlet: HealthCmd= takes a plain command line, not the Docker CMD [...] JSON form:
HealthCmd=/dozzle healthcheck
Older podman-compose (< 1.5.0) runs all healthchecks via sh, which doesn't exist in the Dozzle image. Update to a current version.
Rootless Podman can only access containers in the same user namespace. If running Dozzle as one user, it cannot see containers from another user's rootless session.
Solution: Either run Dozzle as the same user or use rootful mode.