Multi-Container Management

Your server runs 3 AI containers at once (oc-1, oc-2, hm-3). This page covers how to manage them together.

---

The Three Containers

---

Reading Container Status

The MAIN tab’s AI Agents list shows, for each container:

[Screenshot: MAIN tab AI Agents list with all three containers and status]

Click a container name to jump to its tab.

---

Setting Container Nicknames

The default names oc-1, oc-2, hm-3 are machine-friendly but not human-friendly. Set nicknames.

How to Set

1. Open the container tab (Rose / Ada / Vi)
2. Click the edit icon (✏️) next to the title
3. Enter a nickname (under 10 characters recommended)
4. Press Enter to save

[Screenshot: Vi Dashboard title with edit icon and nickname input]

Nickname Suggestions

Once a nickname is set, the original ID still shows as subtext, and the system still uses oc-1 / hm-3 internally.

---

Bulk Start / Stop

From the MAIN tab’s AI Agents list (top-right):

• Start All — boot every stopped container
• Stop All — halt every running container
• Restart All — restart in sequence

Individual actions: from the container tab → Overview → Restart / Stop / Start.

---

Resource Allocation

Each container has independent CPU and memory budgets:

When Resources Run Tight

If the Usage tab shows a container hitting its limits regularly:

1. Check the conversation — did the AI get stuck in a loop or process a huge file?
2. Restart the container — clears transient state
3. Tune the model — switch to a lighter model to reduce consumption
4. Consider upgrading — exceed standard allocation? Contact support.

---

Cross-Container Collaboration

The three containers run independently but can call each other.

Typical Pattern

Ada receives a customer LINE message
    ↓
Triages the content
    ├─ General question → Ada replies
    ├─ Needs system lookup → calls Rose (host root)
    └─ Needs data processing → calls Vi (Hermes sandbox)
         ↓
      Result returned to Ada
         ↓
      Ada synthesizes and replies to the customer

Configuration

Authorize calls to Rose / Vi in Ada’s config, and specify the trigger conditions. See Multi-Agent Workflows.

---

Upgrading Containers Individually

The three containers can be upgraded independently — one does not affect the others.

Rose / Ada (OpenClaw)

Rose Dashboard → Settings → Version Upgrade → Check for Updates → Pick version → Upgrade

Rose and Ada share an OpenClaw version pool; upgrading them together is usually the safe path.

Vi (Hermes)

Vi Dashboard → Settings → Version Upgrade → Check for Updates → Pick version → Upgrade

Vi uses its own Hermes version pool, fully decoupled from Rose and Ada.

---

FAQ

Q: All three containers are down. What now?

Priority order:

1. MAIN tab → Restart Host — reboot the entire server (last resort)
2. Restart one at a time — recover the most critical container first
3. Open a support ticket — include a screenshot of the AI Agents list plus the logs

Q: A container constantly sits at 90%+ CPU — how do I diagnose?

1. Container tab → Activity — what was the last thing it ran?
2. Container tab → Logs — any repeating error messages?
3. Usage tab — is token consumption spiking? (suggests an infinite loop)

Q: Can I add a 4th container?

Not currently. The standard allocation is Rose / Ada / Vi. For more instances:

• Enable Profiles on Vi — one hm-3 container can host multiple isolated instances
• Upgrade your plan — see Subscription Management

---

Related Docs

• MAIN Tab Full Guide
• URL and Token Map
• Multi-Agent Workflows