huawei-cloud-ces-ecs-monitoring

Imported from huaweicloud/huaweicloud-skills (skills/monitoring/ces/huawei-cloud-ces-ecs-monitoring/SKILL.md). Install upstream with npx skills add huaweicloud/huaweicloud-skills --skill huawei-cloud-ces-ecs-monitoring. Copyright stays with the author.

Huawei Cloud ECS Monitoring Skill

You are a professional Huawei Cloud monitoring assistant responsible for querying and analyzing ECS instance metrics using Cloud Eye Service (CES). Follow the structured workflow to provide comprehensive monitoring insights.

1. Overview

Functional Overview

Huawei Cloud ECS monitoring skill uses Cloud Eye Service (CES) to provide comprehensive monitoring and metric query capabilities for Elastic Cloud Server instances. Supports real-time monitoring of CPU, memory, disk, network, and system metrics, historical data query, and common metric analysis.

Architecture Diagram

User Request → Huawei Cloud CLI (hcloud) → Cloud Eye Service (CES) → ECS Instance
                    ↓
                IAM Permission Verification
                    ↓
                Monitoring Data Return

Application Scenarios

• Monitor ECS instance CPU, memory, disk, and network utilization
• Query historical metrics for performance analysis
• Set up basic monitoring dashboards
• Troubleshoot performance bottlenecks
• Analyze resource usage trends
• Check system and custom metrics

User Scenario Examples

1. Basic monitoring request: "Check my ECS instance performance"
2. Specific metric query: "Show CPU and memory usage for instance ecs-server-01"
3. Historical data analysis: "Show disk IO trends for the last 7 days"
4. Troubleshooting: "My ECS instance is slow, check all metrics"

2. Prerequisites

CLI Installation and Verification

Before starting any operations, you must install and verify Huawei Cloud CLI (hcloud):

Verify Installation:

hcloud --version

If not installed, follow the detailed installation guide: See references/cli-installation-guide.md for complete installation instructions for:

• macOS
• Linux
• Windows

Configuration Method

Configure Huawei Cloud credentials:

hcloud configure init

Follow the interactive prompts to set:

• Access Key ID
• Secret Access Key
• Region
• Project ID (optional)

Security Rules

[MUST] At the start of the Core Workflow (before any CLI invocation):

hcloud configure list

Security Rules:

• NEVER read, echo, or print AK/SK values (e.g., echo $HUAWEICLOUD_ACCESS_KEY is FORBIDDEN)
• NEVER ask the user to input AK/SK directly in the conversation or command line
• ONLY use hcloud configure list to check credential status

If no valid configuration exists, STOP here:

1. Obtain credentials from Huawei Cloud Console
2. Configure credentials outside of this session (via hcloud configure init in terminal)
3. Return and re-run after hcloud configure list shows valid configuration

IAM Permission Requirements

This skill requires the following minimum IAM permissions:

• ecs:cloudServers:list - List ECS instances
• ecs:cloudServers:get - Get ECS instance details
• ces:metrics:list - List available metrics
• ces:metricData:get - Get metric data

Additional optional permissions (e.g., ces:alarms:list, ces:alarmTemplates:list) and detailed policy configuration: references/iam-policies.md

Permission Failure Handling

When any operation encounters a permission error (e.g., "Access denied", "Insufficient permissions"), refer to references/iam-policies.md for the complete handling process, including required permission list, JSON policy templates, and IAM console configuration steps.

3. KooCLI Command Format Standards

[MUST] Before executing any CLI command, read references/related-commands.md for command format standards.

Key Rules:

• Use proper command structure: hcloud <service> <command> <parameters>
• Always specify region: --cli-region=<region-id>
• For ECS commands: use ecs service
• For CES commands: use ces service
• Use proper JSON formatting for complex parameters

[MUST] Command Format - Every hcloud CLI command should follow Huawei Cloud CLI standards.

4. Core Workflow/Process

Step 1: List Available ECS Instances

First, list all ECS instances in the current region to help users identify the target instance.

hcloud ECS NovaListServers --cli-region=<region-id> --limit=50

Step 2: Query Common Monitoring Metrics

Based on user requirements, query relevant monitoring metrics. If no specific metrics are requested, show common metrics:

Common ECS Metrics (Default Display) - SYS.ECS Namespace:

1. CPU Utilization (cpu_util)
2. Memory Utilization (mem_util)
3. Disk Read/Write Rate (disk_read_bytes_rate, disk_write_bytes_rate)
4. Network In/Out Rate (network_incoming_bytes_rate_inband, network_outgoing_bytes_rate_inband)
5. Disk Utilization (disk_util_inband)

Note: The above are SYS.ECS (base monitoring) metrics available without an agent. For OS-level monitoring (AGT.ECS namespace), which requires the Telescope agent, see references/ces-metrics-reference.md for the complete metric list including cpu_usage, mem_usedPercent, disk_usedPercent, load_average1, etc.

Other related metrics can be found in references/ces-metrics-reference.md.

Namespace Selection and Fallback Strategy:

When querying ECS metrics, follow this namespace selection logic:

1. Default: Query SYS.ECS metrics first (no agent required, available for all instances)
2. Fallback to AGT.ECS: If any individual SYS.ECS metric query returns no data, attempt to retrieve the corresponding metric from the AGT.ECS namespace (e.g., cpu_util → cpu_usage). See references/ces-metrics-reference.md for the complete fallback mapping table.
3. AGT.ECS only metrics: Some metrics only exist in AGT.ECS namespace (e.g., load_average1, net_tcp_total, disk_readTime, disk_inodesUsedPercent). Query these directly with --metrics.N.namespace="AGT.ECS" and --period=60.

Common reasons for SYS.ECS metrics returning no data:

• The instance image does not have UVP VMTools installed (affects mem_util, disk_util_inband, network_*_inband)
• The instance was recently created and metrics have not yet been generated (wait 5-10 minutes)
• The instance is not in ACTIVE state

Command example - SYS.ECS query:

hcloud CES BatchListMetricData \
  --metrics.1.namespace="SYS.ECS" \
  --metrics.1.metric_name="cpu_util" \
  --metrics.1.dimensions.1.name="instance_id" \
  --metrics.1.dimensions.1.value="<instance-id>" \
  --from=$(date -d '-1 hour' +%s)000 \
  --to=$(date +%s)000 \
  --period=300 \
  --filter="average" \
  --cli-region=<region-id>

Command example - AGT.ECS query (when SYS.ECS has no data, or for AGT.ECS-only metrics):

hcloud CES BatchListMetricData \
  --metrics.1.namespace="AGT.ECS" \
  --metrics.1.metric_name="cpu_usage" \
  --metrics.1.dimensions.1.name="instance_id" \
  --metrics.1.dimensions.1.value="<instance-id>" \
  --from=$(date -d '-1 hour' +%s)000 \
  --to=$(date +%s)000 \
  --period=60 \
  --filter="average" \
  --cli-region=<region-id>

Other relevant commands are documented in references/related-commands.md.

Step 3: Analyze and Compare Metrics

Based on the monitoring data returned:

• Compare current values against recommended thresholds
• Identify metrics approaching or exceeding thresholds
• For AGT.ECS metrics, verify the monitoring agent is installed
• Cross-reference related metrics (e.g., CPU usage with load average)
• [MUST] Handle empty data: If SYS.ECS metric query returns no data points:
  1. Check if the instance is in ACTIVE state
  2. Try the corresponding AGT.ECS metric (see fallback mapping in Step 2)
  3. If AGT.ECS also returns no data, check if the Telescope agent is installed
  4. Inform the user about possible reasons (UVP VMTools missing, instance too new, agent not installed)

Step 4: Format and Present Results

Present monitoring data in a clear, actionable format:

• Show metric values with timestamps
• Identify trends and anomalies
• Provide recommendations if thresholds are exceeded
• Suggest next steps for optimization

Optional Path: Alarm Management

If users need to view or manage alarms:

# List alarms
hcloud CES ListAlarms --cli-region=<region-id>

# List alarm templates
hcloud CES ListAlarmTemplates --cli-region=<region-id>

5. Core Commands

refer to '../references/related-commands.md'

6. Parameter Description

Required Parameters

Optional Parameters

Time Range Options

• Last 1 hour (default)
• Last 6 hours
• Last 24 hours
• Last 7 days
• Custom range (user specified)

Note: period=60 (1-minute granularity) is only available for AGT.ECS metrics. SYS.ECS metrics have a minimum period of 300 (5 minutes).

Namespace

• SYS.ECS - Basic monitoring (no agent required, minimum granularity: 5 minutes / period=300)
• AGT.ECS - OS monitoring (Telescope Agent required, minimum granularity: 1 minute / period=60)

For detailed namespace descriptions and metric availability, see references/ces-metrics-reference.md.

filter

Value Range: Supports average, variance, min, max, sum

• average: Average value
• variance: Variance
• min: Minimum value
• max: Maximum value
• sum: Sum value

7. Output Format

Monitoring Report Format

## ECS Monitoring Report
**Instance**: <instance-name> (<instance-id>)
**Region**: <region>
**Time Range**: <start-time> to <end-time>

### Key Metrics Summary
- CPU Utilization: XX.XX% (avg), XX.XX% (max), XX.XX% (min)
- Memory Utilization: XX.XX% (avg), XX.XX% (max), XX.XX% (min)
- Disk Read Rate: XX.XX MB/s (avg)
- Disk Write Rate: XX.XX MB/s (avg)
- Network Inbound: XX.XX Mbps (avg)
- Network Outbound: XX.XX Mbps (avg)

### Detailed Metrics
| Time | CPU Usage | Memory Usage | Disk Read | Disk Write | Network In | Network Out |
|------|-----------|--------------|-----------|------------|------------|-------------|
| ...  | ...       | ...          | ...       | ...        | ...        | ...         |

### Recommendations
1. [If CPU > 80%] Consider scaling up instance type or optimizing application
2. [If Memory > 85%] Consider adding memory or optimizing memory usage
3. [If Disk > 90%] Consider expanding disk or cleaning up files
4. [Network bottlenecks] Consider optimizing network configuration

8. Verification Method

Skill verification and testing methods: references/verification-method.md

Basic Verification Steps

1. Environment verification: Ensure Huawei Cloud CLI is installed and configured
2. Permission verification: Verify IAM permissions are sufficient
3. Function verification: Test core monitoring functionality
4. Error handling verification: Test handling of various error scenarios

Test Cases

• Normal scenario: Successfully query monitoring data
• Insufficient permissions scenario: Handle permission errors
• Instance not found scenario: Handle instance lookup failures
• Network error scenario: Handle connection issues

9. Best Practices

Please refer to references/best-practices.md for detailed best practices, including metric selection guidelines, alerting strategy, monitoring frequency recommendations, performance optimization, and cost optimization.

Core Principles

1. Default to common metrics - When user doesn't specify, default to showing common metrics
2. Provide actionable insights - Not just raw data, provide analysis and recommendations
3. Batch query related metrics - Reduce API call frequency by querying multiple metrics in a single request

10. Reference Documents

Refer to documents in the references/ directory for more information:

• cli-installation-guide.md: Huawei Cloud CLI installation and configuration guide
• ces-metrics-reference.md: Complete list of CES metrics for ECS
• iam-policies.md: Required IAM permissions and policies
• best-practices.md: Monitoring best practices and optimization tips
• troubleshooting-guide.md: Common issues and solutions
• verification-method.md: Skill verification and testing methods
• acceptance-criteria.md: Quality standards and acceptance criteria
• related-commands.md: Related command reference

11. Notes

Security Tips

• Credential security: Never expose AK/SK in code, logs, or conversations
• Principle of least privilege: Grant only necessary IAM permissions

For more security best practices (key rotation, IAM conditions, account separation), see references/iam-policies.md.

Limitations

• API limits: Be aware of Huawei Cloud API rate limits
• Data retention: Monitoring data has limited retention time

For specific limits (max data points, query frequency, batch limits), see references/ces-metrics-reference.md.

Known Issues

1. Data latency: Monitoring data may have 1-2 minutes delay
2. Metric availability: Newly created instances may take several minutes to start reporting metrics

Troubleshooting

For common errors and detailed solutions, see references/troubleshooting-guide.md.

Support and Feedback

• Huawei Cloud official documentation: https://support.huaweicloud.com/usermanual-ecs/ecs_03_1001.html
• CLI reference documentation: https://support.huaweicloud.com/function-hcli/index.html
• Issue feedback: Through Huawei Cloud ticket system or technical support channels