Podman keeps being enable on pulse-agent even when installed with --disable-docker

[wilddev65]

Describe the bug
Due to changing pulse to use TLS, I reinstalled the pulse-agent on my Rocky Linux 9.1 system. After this, the agent re-enabled detection of Podman on the system.

To Reproduce
Steps to reproduce the behavior:

1. Uninstall current agent with --uninstall
2. Remove host from Agents in settings.
3. Validate host is not on Hosts or Docker screen
4. Restart Pulse LXC just to make sure
5. Reinstall agent using the following command:
   curl -fsSL https://pulse:7655/install.sh | bash -s -- --url https://pulse:7655 --token --interval 30s --disable-docker
6. Check log with journalctl -u pulse-agent | grep docker

Jan 22 17:24:59 mani pulse-agent[8100]: {"level":"info","version":"v5.0.17","pulse_url":"https://pulse:7655","host_agent":true,"docker_agent":false,"kubernetes_agent":false,"proxmox_mode":false,"auto_update":true,"time":"2026-01-22T17:24:59-07:00","message":"Starting Pulse Unified Agent"}
Jan 22 17:25:00 mani pulse-agent[8100]: {"level":"warn","component":"pulse-docker-agent","time":"2026-01-22T17:25:00-07:00","message":"Podman runtime detected; disabling Swarm service collection"}
Jan 22 17:25:00 mani pulse-agent[8100]: {"level":"warn","component":"pulse-docker-agent","time":"2026-01-22T17:25:00-07:00","message":"Podman runtime detected; disabling Swarm task collection"}
Jan 22 17:25:00 mani pulse-agent[8100]: {"level":"info","component":"pulse-docker-agent","runtime":"podman","daemon_host":"unix:///run/user/0/podman/podman.sock","version":"5.6.0","time":"2026-01-22T17:25:00-07:00","message":"Connected to container runtime"}

Expected behavior
I expected to not see the agent for this host in the Docker list.

Environment:

• Pulse Version: v5.0.17
• Installation Type: Proxmox VE LXC

Additional context

[rcourtman]

Found the root cause! The install script's systemd service configuration was missing the explicit --enable-docker=false flag when --disable-docker was used.

What happened: When you installed with --disable-docker, the agent binary wasn't receiving the explicit --enable-docker=false argument. Without that flag, the agent's auto-detection logic saw Podman available and enabled Docker monitoring anyway.

Pushed a fix that ensures the --enable-docker=false flag is properly passed to the systemd service file when Docker monitoring is explicitly disabled during installation.

Workaround until next release: You can manually edit your systemd service file to add the flag:

sudo systemctl stop pulse-agent
sudo sed -i 's/--enable-host/--enable-docker=false --enable-host/' /etc/systemd/system/pulse-agent.service
sudo systemctl daemon-reload
sudo systemctl start pulse-agent

Or set the environment variable:

# Add to /etc/systemd/system/pulse-agent.service under [Service]
Environment="PULSE_ENABLE_DOCKER=false"

---

Automated triage.

[wilddev65]

I edited the system service file like you said above. This results in the docker host showing offline:

Not sure if that is expected behavior. It does seem to have taken from what the log messages are saying:
Jan 26 13:51:41 mani pulse-agent[113386]: {"level":"info","version":"v5.0.17","pulse_url":"https://pulse:7655","host_agent":true,"docker_agent":false,"kubernetes_agent":false,"proxmox_mode":false,"auto_update":true,"time":"2026-01-26T13:51:41-07:00","message":"Starting Pulse Unified Agent"}

In addition, the agent list in settings indicates Docker as a tag. Thanks for your work on this!

[rcourtman]

The offline status for the Podman host is expected when Docker/Podman monitoring is disabled - since the agent is no longer sending container data, Pulse marks the host as offline (it hasn't "checked in" with container info recently).

To clean it up, you can remove the stale entry from the UI:
Settings → Agents → Docker Hosts → find "mani" and click the delete/remove button.

This will remove the entry from your dashboard. The Docker "tag" in the agent list should also clear after this.

The underlying install script fix is in place, so future installations with --disable-docker will work correctly without creating container host entries in the first place.

---

Automated triage.

[rcourtman]

Addressed — when Docker/Podman monitoring is disabled, the offline status is expected since the agent stops sending container data. The dashboard now reflects this correctly.