Network Topology

Use Network Topology for real-time visibility into physical and virtual network health across compute nodes.

This page combines host-level and fabric-level networking views in one place:

• Dashboard summary cards for overall health and capacity indicators

• Nodes tab for per-node quick status

• Alerts tab for active network issues

• Node detail tabs for Bridges, Interfaces, VXLANs, Bonds, VLAN Interfaces, and Routes

Term Definitions:

• MTU: Maximum Transmission Unit, the largest packet size sent without fragmentation.

• Carrier: whether a network link is physically/operationally up.

• RX/TX: received/transmitted traffic counters.

• CRC: Cyclic Redundancy Check error indicator for link/data corruption.

Step: Open Network Topology

When to Use: Use this for initial network-fabric triage and node-level drilldown.

Purpose: Open Network Topology and establish a baseline before reviewing nodes or alerts.

Dashboard shows:

• Summary cards for network health, total interfaces, total bridges, total errors, MTU issues, and total traffic.

• Nodes and Alerts tabs for host-level and issue-level troubleshooting entry points. Both tab labels carry a live count, for example Nodes (6) / Alerts (2).

• Nodes is the default first view and shows the Host Table immediately on page load.

• Node rows show physical link state, top talker, offloads, anchor bridge, bridge inventory, active VXLAN count, and an action to open details.

What you can do from this screen:

• Start triage from high-level health indicators.

• Switch to Alerts for active issues.

• Open node details for bridges, interfaces, and VXLAN analysis.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. Confirm Nodes is selected (default first view).

3. Review dashboard summary cards.

4. Review the node list shown in the first view.

5. Use Nodes and Alerts tabs for triage.

6. Click a node to open detail tabs.

Expected Outcome:

• You get the Nodes-first dashboard view with summary cards and node list visible.

• You get direct entry points for alert review and node-level troubleshooting.

If this fails:

1. Reload Network Topology and clear any active node search/filter before retrying.

2. Confirm your account can access Network Topology.

3. Confirm networking telemetry is available for your environment.

Step: Open Network Topology Help Panel

When to Use: Use this after opening the Network Topology dashboard when you need definitions for cards, tabs, and alerts.

Purpose: Open page help as a separate step from dashboard review.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. Click the top-right help icon.

3. Review the help panel guidance.

Expected Outcome:

• Network Topology help opens in the right-side panel.

• You can review the complete reference for dashboard cards, node details, and tab-level troubleshooting.

• You can continue dashboard triage with clear context.

If this fails:

1. Refresh the page and click the help icon again.

2. Confirm your account can access in-product help content.

3. Retry after reopening Network Topology.

Summary Cards

Card	Description	Click Action	Common Task
Network Health	Overall health percentage across nodes and network objects.	Informational	Monitor overall networking baseline.
Total Interfaces	Count of links up versus total interfaces.	Informational	Investigate when links are down.
Total Bridges	Total Linux bridge count across nodes.	Informational	Validate bridge topology scale.
Total Errors	Cumulative NIC/bridge error count.	Open Alerts tab	Drill into active error conditions.
MTU Issues	Nodes with MTU mismatch/drift conditions.	Informational	Use Alerts tab to identify affected devices.
Total Traffic	Aggregate traffic volume across nodes.	Informational	Capacity and traffic trend review.

Warning

MTU mismatches between bridges and interfaces can cause packet drops and unstable connectivity.

Step: Review Nodes View (Default)

When to Use: Use this to identify which host requires detailed bridge/interface/VXLAN investigation.

Purpose: Review node status and open a specific node for deeper network analysis.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. Stay on the default Nodes view.

3. Review the host list.

4. Review physical link, top talker, offload, bridge, and VXLAN summary fields.

5. Use the row action to open node details.

Expected Outcome:

• You can identify which node requires deeper bridge/interface/VXLAN inspection.

Nodes table:

Column	Description
Host / Node	Host IP or FQDN. Click to open node detail page.
Physical Link	Primary physical link state and interface name.
Top Talker	Highest-traffic interface or bridge on that node.
Offloads	LRO/GRO/TSO status summary.
Anchor Bridge	Primary bridge and MTU shown for the node.
Bridges	L2 bridge membership and MTU status.
Active VXLANs	Number of attached VXLAN interfaces.
Actions	Open node details.

If this fails: 1. Refresh Nodes and clear any active filter. 2. Confirm node inventory is available in your environment. 3. Reopen Network Topology and retry node selection.

Step: Review Alerts Tab

When to Use: Use this when troubleshooting active network faults by severity and category.

Purpose: Review current network alerts and pivot to affected nodes for root-cause analysis.

Steps:

1. In Network Topology, click Alerts.

2. Filter by node when focusing investigation.

3. Open the affected node from the alert row.

4. Verify issue context in node detail tabs.

Expected Outcome:

• You can isolate active network faults by severity/category and pivot directly to the affected node.

If this fails:

1. Reopen the Alerts tab and confirm alert stream data is loading.

2. Confirm alert stream and node inventory are available.

3. Reopen Network Topology and retry Alerts tab.

Network Alerts table:

Severity	Category	Node	Device	Message
Critical/Warning/Info	Link Down, Errors, MTU, MTU Drift, node_offline, ntp, vxlan_fabric	Host where alert occurred	Interface/bridge name	Human-readable issue description

Alert categories:

• Link Down: interface operational/admin down

• Errors: elevated RX/TX/CRC errors

• MTU: MTU mismatch between connected components on one node

• MTU Drift: cluster-wide - the same bridge name carries different MTUs on different nodes (for example Cluster MTU drift: MTU 1500 on node-a, MTU 9000 on node-b)

• node_offline: the node is unreachable, so no per-device data was collected (Node unreachable)

• ntp: the node’s clock is not synchronised, or its offset has drifted past the warning threshold

• vxlan_fabric: cluster-wide - only some online KVM nodes have active VXLAN tunnels; cross-check on SDN Fabric

Note

The last three categories are shown as raw identifiers (node_offline, ntp, vxlan_fabric) rather than friendly names. node_offline rows have no Device; vxlan_fabric is cluster-wide, so it has neither Node nor Device.

Node Detail Page

Click any node from Nodes or Alerts to open node details.

The node detail view includes six operational tabs:

• Bridges

• Interfaces

• VXLANs

• Bonds

• VLAN Interfaces

• Routes

Bridges Tab

When to Use:

Use this when network issues point to bridge state, bridge membership, carrier loss, or bridge-level errors.

Purpose:

Validate Linux bridge health and identify which bridge/member is causing impact.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. In Nodes, click the target node to open node details.

3. Open the Bridges tab.

4. Review cards and bridge rows for state, MTU, members, carrier, and error indicators.

Expected Outcome:

• You can confirm whether bridge-layer connectivity is healthy.

• You can identify the affected bridge/member for follow-up action.

If this fails:

1. Refresh node details and reopen Bridges.

2. Confirm the node is online in Nodes.

3. Retry from the related Alerts row for the same node.

Use this tab to inspect Linux bridge state, membership, and error counters.

Bridge details page:

Bridge stats cards include:

• Total bridges

• Active

• Inactive/Down

• Total errors

• Most traffic

Bridge row details include:

• Name (for example cloudbr0)

• Layer type and MAC address

• Binding/member interface

• Bridge MTU and TX queue

• Admin, operational, and carrier state

• Flags and member interfaces

• Traffic, error, and drop counters

• Expandable QoS details and a QOS PROFILE field (for example Aqm / fq_codel)

• Member interfaces chips - click one to jump straight to that interface

Managing Bridges

• + Create Bridge opens a form to add a new Linux bridge on the node.

• Platform-managed bridges (badge Platform Managed) cannot be edited or deleted from here - they are owned by traffic-type/fabric configuration.

• A bridge you created yourself exposes edit and delete actions on its row.

Interfaces Tab

When to Use:

Use this when you need NIC-level or virtual-interface-level status and packet counters.

Purpose:

Verify interface state, drops, and errors to isolate link or port-level faults.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. In Nodes, click the target node to open node details.

3. Open the Interfaces tab.

4. Review links-up, error, and drop counters and check affected interface rows.

Expected Outcome:

• You can identify which interface is down, noisy, or dropping traffic.

• You can separate interface faults from bridge or overlay faults.

If this fails:

1. Refresh node details and reopen Interfaces.

2. Confirm interface telemetry is available for the node.

3. Cross-check with Alerts for the same node/device.

Use this tab to inspect physical and virtual interface health and counters.

Interface stats cards include:

• Links up

• Packet errors

• Packet drops

• VXLAN tunnels

Interface table fields include:

• Interface

• Layer and type

• Master

• MTU

• TX queue

• Admin, operational, and carrier state

• RX/TX traffic

• State change age

• Hardware details (driver, bus, vendor) and linked bridge

• Expandable QoS details and a QOS PROFILE field (for example Aqm)

Managing Interfaces

• + Create Interface adds a new interface on the node.

• Platform-managed interfaces carry a locked Platform Managed badge. They cannot be deleted, and the admin up/down toggle and MTU are locked in the edit panel - changing either can isolate the node or break the fabric path MTU. They are not fully read-only: offloads, RX ring / TX ring buffers, and TX queue length stay editable. On the platform-owned public-egress interface, QoS is editable as well.

• A non-platform-managed interface (for example a manually added dummy/test interface) shows edit (pencil) and delete (trash) actions on its row.

• Some interfaces are fully read-only and offer no actions: system virtual devices (loopback and virtualization plumbing) and interfaces of unattributed origin. They carry a Read-only badge, and the platform refuses mutations on them server-side.

VXLANs Tab

When to Use:

Use this when overlays are suspected (cross-host VM connectivity, VNI spread, tunnel state).

Purpose:

Validate VXLAN tunnel health and VNI participation on the selected node.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. In Nodes, click the target node to open node details.

3. Open the VXLANs tab.

4. Review tunnel state, VNI values, local/remote endpoint fields, and error/drop counters.

Expected Outcome:

• You can confirm whether VXLAN tunnels are up and VNIs are present as expected.

• You can identify overlay mismatches for targeted SDN follow-up.

If this fails:

1. Refresh node details and reopen VXLANs.

2. Verify the node has active VXLAN interfaces in this environment.

3. Cross-check with SDN Fabric EVPN/VNI views.

Use this tab to inspect overlay tunnels and VNI participation state.

VXLAN stats cards include:

• VXLAN tunnels

• Tunnels up

• Unique VNIs

• Learning off

VXLAN card fields include:

• Name

• VNI

• Local IP

• Group/multicast value

• MTU and TX queue

• Learning state

• Admin and operational state

• Platform Managed badge - present when the fabric (not you) created the tunnel

Bonds Tab

Lists any bonded (LACP/active-backup) interfaces on the node. Empty on a node with no bonds configured. Bond rows follow the same state/MTU/carrier fields as Interfaces.

VLAN Interfaces Tab

When to Use:

Use this to inspect or create 802.1Q VLAN sub-interfaces on a node’s physical NICs.

Steps:

1. Open Control Center -> Network -> Network Topology.

2. In Nodes, click the target node to open node details.

3. Open the VLAN Interfaces tab.

Stats cards show VLAN Interfaces, Healthy, Degraded, and Down counts. Each row shows the sub-interface (for example eth4.3115), VLAN tag, Parent NIC, Master Bridge, MTU, admin/operational/carrier state, a QoS details expander, and a Path breadcrumb (parent NIC -> VLAN tag -> bridge) - click any segment to jump to that resource.

Create a VLAN Interface

Before you start:

• The Parent NIC list only offers interfaces that can actually parent a VLAN: it excludes existing VLAN sub-interfaces (VLAN-on-VLAN is not supported), bridges, VXLAN devices, dummy devices, and loopback, and it excludes any NIC that is not both admin and operationally up. If a NIC you expect is missing, check its state on the Interfaces tab first.

• The list is populated only after the target node resolves. Create stays disabled until then.

• Creating the interface requires the network-management permission.

Steps:

1. Click + Create VLAN Interface.

2. Select the Parent NIC and enter a VLAN ID (1-4094).

3. Optionally set MTU. It defaults to the parent’s MTU and cannot exceed it.

4. Choose an IP Configuration:

   • Manual (no IP - L2 only) - the default. Creates the sub-interface with no address, for use as a bridge member or tagged L2 path.

   • Static - reveals Address (CIDR), Gateway, and DNS servers.

   • DHCP - takes no further input.

5. Click Create.

Note

A Gateway entered here is applied at the node’s next boot, not live. The platform deliberately avoids changing a running node’s default route. Do not expect the new gateway to take effect immediately.

If this fails:

1. Static IP validation is unavailable - the Karios Atlas DCIM tool is still initializing. The static-address collision check fails closed rather than skipping. Retry in a few seconds; do not work around it by switching to Manual.

2. IP ... is already in use - the static address is already registered elsewhere. Pick a different address.

3. VLAN sub-interface already exists - that Parent NIC and VLAN ID pair is already present on the node. Check the list before recreating it.

4. If the request is rejected as misdirected, reopen the node detail page from Nodes so the request is routed to the correct node, then retry.

Routes Tab

Shows the same live kernel routing table (FIB) as the Kernel Routes tab under SDN Fabric node details - Protocol, Destination, Gateway, Device, Scope, Pref. Source, Metric, and Flags - reached here from the Network Topology side instead. + Create route opens the same Create kernel route form (Destination (CIDR), Gateway, Device, Metric) described there.

Common Tasks

• Check network health: start from Network Health, then move to Alerts when degraded.

• Investigate errors: open Total Errors and filter by node/device.

• Inspect a node: open node details and compare Bridges, Interfaces, VXLANs, Bonds, VLAN Interfaces, and Routes.

• Find MTU issues: use MTU Issues and Alerts categories MTU or MTU Drift.

• Verify overlay status: in VXLANs, confirm Tunnels Up aligns with expected tunnel count.

---

→ Next: IPAM