Perform batch inspection on Alibaba Cloud MSE Nacos instances, checking configuration count usage, connection count usage, QPS usage, and TPS usage. Supports two modes: inspecting a specified list of instance IDs, or inspecting all Nacos instances in a region by region. Triggers when the user needs to perform health inspection, capacity assessment, or usage check on MSE Registration & Configuration Center instances, or when the user mentions "inspect nacos", "check nacos instance usage", "nacos capacity inspection", "MSE registry inspection", "nacos inspection", "check nacos config count", "nacos connection count over limit", "nacos QPS/TPS usage", "inspect all nacos in a region", "inspect nacos instances in cn-hangzhou", "check nacos usage across all regions"
Perform usage inspection on a specified set of Alibaba Cloud MSE Nacos instances, covering configuration count usage, connection count usage, QPS usage, and TPS usage, and output alert recommendations based on thresholds. Supports two inspection modes: inspecting a specified list of instance IDs, or automatically discovering and inspecting all Nacos instances in a region.
Pre-check: Aliyun CLI >= 3.3.3 required
Run aliyun version to verify >= 3.3.3. If not installed or version too low,
or see references/cli-installation-guide.md for installation instructions.
Pre-check: Aliyun CLI plugin update required
[MUST] run aliyun configure set --auto-plugin-install true to enable automatic plugin installation.
[MUST] run aliyun plugin update to ensure that any existing plugins are always up-to-date.
Observability (MUST follow for every aliyun command)
Upon loading this skill, generate a random session ID (32-char lowercase hex string) once
for the entire session. Use it as {session-id} below.
Rule: Every aliyun CLI command that calls a cloud API MUST include the --user-agent flag.
Local utility commands (e.g. configure, plugin, version) do not support this flag
and should be excluded
mse:ListClusters - List all MSE instances in a region (used in region-based inspection mode)
arms:ListPrometheusInstances - List Prometheus instances to obtain cloud product Prometheus instance IDs.
arms:GetPrometheusInstance - Get Prometheus HTTP API public and internal addresses
For the full permission policy, see references/ram-policies.md.
[MUST] Permission Failure Handling: When any command or API call fails due to permission errors at any point during execution, follow this process:
Read references/ram-policies.md to get the full list of permissions required by this SKILL
Use ram-permission-diagnose skill to guide the user through requesting the necessary permissions
Pause and wait until the user confirms that the required permissions have been granted
Workflow
Step 1: Collect Inspection Information
IMPORTANT: Parameter Confirmation — Before executing any command or API call,
ALL user-customizable parameters (e.g., RegionId, instance names, CIDr blocks,
passwords, domain names, resource specifications, etc.) MUST be confirmed with the
user. Do NOT assume or use default values without explicit user approval.
Before starting the inspection, proactively confirm the following information with the user (if already provided, use directly):
Parameter
Required/Optional
Description
Default
Inspection Mode
Required
Mode A: Specify instance ID list; Mode B: Inspect by region
None
Instance ID List
Required for Mode A
Specific MSE instance IDs (e.g., mse-cn-xxx1, mse-cn-xxx2)
None
Region
Required
Region of the instances or target region for region-based inspection (e.g., cn-hangzhou)
None
Inspection Time Range
Optional
Time range for monitoring metric queries
Last 1 hour
Inspection Mode — Confirm which mode the user wants to use:
Mode A: Specify Instance List — The user provides specific instance IDs (e.g., mse-cn-xxx1, mse-cn-xxx2) along with the region for each instance
Mode B: Inspect by Region — The user provides one or more regions (e.g., cn-hangzhou, cn-shanghai), and the system automatically discovers and inspects all Nacos instances in those regions
Inspection Time Range — For example, "last 1 hour", "last 30 minutes", etc. Default is last 1 hour
Decision criteria: If the user provides specific instance IDs, use Mode A; if the user only mentions a region without specifying instance IDs (e.g., "inspect all nacos instances in cn-hangzhou"), use Mode B.
Step 2: Get the List of Instances to Inspect
Mode A: User Has Specified Instance IDs
Use the instance ID list provided by the user directly, proceed to Step 3 to query details for each instance.
Mode B: Auto-discover All Instances by Region
For each region specified by the user, list all MSE instances in that region via paginated queries.
Execute the query and extract the instance list from the response
Extract InstanceId, ClusterName, MseVersion and other fields from each instance record.
If the number of instances returned on the current page equals PageSize, increment PageNum by 1 and query the next page.
If the current page returns an empty instance list or fewer instances than PageSize, all instances have been retrieved; stop paginating
Important: The list-clusters response already includes ClusterName and MseVersion fields. Therefore, for successfully listed instances in Mode B, there is no need to call query-cluster-detail separately — you can use the field values from the list response directly, skip Step 3, and proceed to Step 4 to query Prometheus metrics. If list-clusters returns an empty list, the inspection should be paused while simultaneously asking the customer if the region is correct.
Step 3: Query Instance Details (Mode A Only)
For each Nacos instance to be inspected in Mode A, execute the following command to get ClusterName and MseVersion:
Recommended step parameter: use 15s for inspection time range <= 1 hour, 60s for within 1 day.
For each instance, query the following metrics based on its edition type. Replace $clusterName in all PromQL queries with the ClusterName value obtained in Step 2 (Mode B) or Step 3 (Mode A) for that instance.
For complete PromQL queries, refer to references/promql-queries.md.
4.1 Configuration Count Usage
Professional (mse_pro): Query references/promql-queries.md §4.1 Professional
For each instance and each metric, extract the maximum value from the time series data returned by the range query as the basis for inspection judgment.
5.2 Alert Threshold Determination
Metric
Alert Threshold
Alert Action
Configuration count usage
> 0.8 (80%)
Alert user to monitor configuration count quota and adjust in time
Connection count usage
> 1 (100%)
Alert user to monitor instance capacity and upgrade in time
QPS usage
> 1 (100%)
Alert user to monitor instance capacity and upgrade in time
TPS usage
> 1 (100%)
Alert user to monitor instance capacity and upgrade in time
5.3 Output Format
Summarize the inspection results in a Markdown table in the session, and also generate a .md inspection report file.
Session Output Example:
text
## MSE Nacos Inspection Results
Inspection Period: 2024-01-01 10:00 ~ 2024-01-01 11:00
Number of Instances Inspected: 3
| Instance ID | Edition | Config Count Usage | Connection Count Usage | QPS Usage | TPS Usage | Status |
|-------------|---------|-------------------|----------------------|-----------|-----------|--------|
| mse-cn-xxx1 | Professional | 45% | 60% | 30% | 20% | Normal |
| mse-cn-xxx2 | Enterprise | 85% | 50% | 40% | 25% | ⚠️ Config count usage over threshold |
| mse-cn-xxx3 | Serverless | 30% | - | - | - | Normal |
### Alert Details
⚠️ **mse-cn-xxx2 (Enterprise)**
- Configuration count usage reached 85%, exceeding the 80% threshold
- Recommendation: Please adjust the configuration count quota in time
- Reference: https://help.aliyun.com/zh/mse/product-overview/mse-quotas-and-limits
5.4 Reference Documentation
Include the following reference documentation links in alert messages:
After the inspection is complete, follow the steps in references/verification-method.md to verify the correctness of the results.
Key verification points:
All four usage metrics have been queried for each instance (or marked as "N/A" / "No data")
Instances exceeding thresholds include alert details and recommendations
Inspection report file has been generated
Cleanup
This skill performs read-only operations only — it does not create, modify, or delete any cloud resources; no cleanup is needed.
Local .md inspection report files generated during the inspection are managed by the user.
Error Handling
No instances in region: If list-clusters returns an empty list for a region, inform the user that there are no MSE Nacos instances in that region and continue inspecting other regions
Pagination query failure: If a page query fails during list-clusters pagination, record the successfully retrieved instance list, continue inspecting the retrieved instances, and note the pagination interruption in the report
Instance does not exist or query failure: If query-cluster-detail returns an error, record the reason for the failure, continue inspecting the remaining instances, and note it in the final inspection report
Insufficient permissions: If AccessDenied or NoPermission is returned, prompt the user to add the corresponding RAM permissions
Prometheus query returns no data: If a metric query returns empty results, mark it as "No data" in the inspection results and suggest the user confirm whether the instance is running normally
Partial instance inspection failure: Does not affect the inspection of other instances; mark failed instances and reasons in the final summary
Prometheus instance does not exist: If list-prometheus-instances returns an empty list, inform the user that no cloud product Prometheus instance has been created in that region and suggest enabling it in the ARMS console
API address is empty: If both HttpApiInterUrl and HttpApiIntraUrl are empty, the instance may not be fully initialized; if the public address is empty but the internal address exists, use the internal address
PromQL syntax error: If Prometheus returns a query error (e.g., errorType: "bad_data"), display the error message and suggest checking the PromQL syntax
Query timeout: If a timeout error is returned, suggest narrowing the query time range or simplifying the PromQL statement
Notes
This skill performs read-only operations only; no cloud resources will be modified
Prometheus queries are executed via the public address (HttpApiInterUrl) by default; no VPC internal network is required; if public access is unavailable, it will automatically fall back to the internal address (HttpApiIntraUrl)
Recommended query step parameter: use 15s for inspection time range <= 1 hour, 60s for within 1 day
For batch inspection of a large number of instances (>10), it is recommended to execute in batches to avoid API rate limiting
When inspecting by region, ListClusters defaults to 10 items per page and will automatically paginate until all instances are retrieved
Developer and Basic edition instances do not support Prometheus monitoring and will be marked as "N/A" in the inspection results
