Node-Level Cluster Monitoring

Overview

Yeedu clusters now support node-level monitoring and management. Monitor individual cluster nodes with detailed metrics, view node health, and track node-specific events.

Node-Level Dashboard

Accessing Node Dashboard

1. Navigate to Clusters
2. Click on a specific cluster
3. Click Nodes tab
4. See all nodes in the cluster with status

Node Information

When you click on a specific node, the node details panel shows:

• Node name/ID: Specific node identifier
• Status: Current node status such as ERROR, STARTING, DESTROYING, or DESTROYED
• Start Time: When the current node lifecycle event began
• End Time: When the node finished or transitioned states
• Duration: Duration of the current or most recent lifecycle operation
• Logs: Quick link to view node logs for troubleshooting

Node Metrics

Real-Time Metrics

For each running node, monitor:

Metric	Description	Range
CPU Usage	Percentage of CPU utilized	0-100%
Memory Usage	GB of memory in use	Absolute value
Disk Usage	GB disk space used	Absolute value
Network In	Incoming network bandwidth	Mbps
Network Out	Outgoing network bandwidth	Mbps
Uptime	How long node has been running	Duration

Node Health Status

Status Indicators

• ✓ Running: Node is operational
• ⊘ Stopping: Node is being shut down
• ⚠ Error: Node encountered an error
• ⊗ Destroyed: Node has been removed
• ? Starting: Node is starting up

Health Checks

Node health determined by:

• Heartbeat from node
• Agent connectivity
• Resource availability
• System processes status

Node Issues

If node shows errors:

1. Click on node details
2. View Error Details section
3. Review error message
4. Check Recent Events
5. Troubleshoot based on reason

Node Events

View all events for the cluster:

1. Click Events tab in cluster
2. Then click on Nodes tab for Node Events: Individual node operations

Node Operations

View Node Details

1. In the Nodes tab, click on a node name
2. See the node details panel with:
   • Node name/ID and role (Worker or Master)
   • Current status and lifecycle information
   • Start time, end time, and duration for the node event
   • A link to view node logs
   • Hardware specifications, network configuration, and resource allocation when available

Node Resource Allocation

View resources allocated to node:

• vCPU: Number of cores
• Memory: RAM amount
• Storage: Local disk capacity
• GPU (if applicable): GPU resources

Node Health and Reliability

Monitor node health across CPU, memory, disk, and network to keep the cluster stable.

• CPU: Normal 20-80%, warning above 85%, critical above 95%.
• Memory: Normal 30-80%, warning above 85%, critical above 95%.
• Disk: Normal 20-70%, warning above 80%, critical above 90%.
• Network: Normal varies by workload; warning when sustained high bandwidth or timeouts occur.

When resources are high:

• Check running jobs and resource usage
• Review logs and process activity
• Optimize workloads and data locality
• Scale or adjust cluster capacity where needed

Yeedu also supports node reliability controls:

• Destroying a master node typically triggers worker destruction to prevent orphaned nodes.
• Enable Master Protection in cluster settings to require confirmation before master destruction.