elasticsearch_health

Instructions

Implementation Reference

• src/tools/getClusterHealth.ts:3-50 (handler)

  The getClusterHealth function executes the tool logic: calls esClient.cluster.health() with optional index-level detail, formats and returns cluster health status as text content.

  export async function getClusterHealth(
    esClient: Client,
    includeIndices: boolean = false
  ) {
    try {
      const response = await esClient.cluster.health({
        level: includeIndices ? "indices" : "cluster"
      });
  
      const content: { type: "text"; text: string }[] = [];
  
      // 添加集群状态概述
      content.push({
        type: "text" as const,
        text: `Cluster Name: ${response.cluster_name}\nStatus: ${response.status}\nNodes: ${response.number_of_nodes}\nData Nodes: ${response.number_of_data_nodes}\nActive Shards: ${response.active_shards}\nActive Primary Shards: ${response.active_primary_shards}\nRelocating Shards: ${response.relocating_shards}\nInitializing Shards: ${response.initializing_shards}\nUnassigned Shards: ${response.unassigned_shards}\nPending Tasks: ${response.number_of_pending_tasks}\n`
      });
  
      // 如果请求了索引级别的健康状态
      if (includeIndices && response.indices) {
        const indicesHealth: string[] = [];
        
        for (const [indexName, indexHealth] of Object.entries(response.indices)) {
          indicesHealth.push(`Index: ${indexName}\n  Status: ${indexHealth.status}\n  Primary Shards: ${indexHealth.number_of_shards}\n  Replicas: ${indexHealth.number_of_replicas}\n  Active Shards: ${indexHealth.active_shards}\n  Active Primary Shards: ${indexHealth.active_primary_shards}\n  Unassigned Shards: ${indexHealth.unassigned_shards}`);
        }
  
        if (indicesHealth.length > 0) {
          content.push({
            type: "text" as const,
            text: `\nIndices Health Status:\n${indicesHealth.join('\n\n')}`
          });
        }
      }
  
      return {
        content
      };
    } catch (error) {
      console.error(`获取集群健康状态失败: ${error instanceof Error ? error.message : String(error)}`);
      return {
        content: [
          {
            type: "text" as const,
            text: `错误: ${error instanceof Error ? error.message : String(error)}`
          }
        ]
      };
    }
  }

• src/server.ts:106-120 (registration)

  The tool registration on the MCP server: defines the name "elasticsearch_health", its description, the Zod schema for the includeIndices parameter, and the handler callback that delegates to getClusterHealth.

  // Get the health status of the Elasticsearch cluster, optionally include index-level details
  server.tool(
    "elasticsearch_health",
    "Get the health status of the Elasticsearch cluster, optionally include index-level details",
    {
      includeIndices: z
        .boolean()
        .optional()
        .default(false)
        .describe("Whether to include index-level details"),
    },
    async ({ includeIndices }) => {
      return await getClusterHealth(esClient, includeIndices);
    }
  );

• src/server.ts:110-116 (schema)

  The input schema for the tool: a single optional boolean field 'includeIndices' with a default of false, validated using Zod.

  {
    includeIndices: z
      .boolean()
      .optional()
      .default(false)
      .describe("Whether to include index-level details"),
  },

• src/tools/getClusterHealth.ts:1-1 (helper)

  The import of the Elasticsearch Client dependency used by the handler.

  import { Client } from "@elastic/elasticsearch";