在K8s集群通过Helm部署Elasticsearch时遇GeoLite2数据库获取失败问题

AI 小北

2026-7-21

问题描述

在AWS节点搭建的K8s集群上，通过Helm执行命令 helm install elasticsearch elastic/elasticsearch -f values-aws.yaml 部署elasticsearch-8.5.1集群后，出现如下错误日志：

{"@timestamp":"2023-05-16T01:22:09.414Z", "log.level":"ERROR", "message":"failed to retrieve database [GeoLite2-City.mmdb]", "ecs.version": "1.2.0","service.name":"ES_ECS","event.dataset":"elasticsearch.server","process.thread.name":"elasticsearch[elasticsearch-master-0][generic][T#2]","log.logger":"org.elasticsearch.ingest.geoip.DatabaseNodeService","elasticsearch.node.name":"elasticsearch-master-0","elasticsearch.cluster.name":"elasticsearch","error.type":"org.elasticsearch.cluster.block.ClusterBlockException","error.message":"blocked by: [SERVICE_UNAVAILABLE/1/state not recovered / initialized];"}

使用的values-aws.yaml配置内容如下：

# Shrink default JVM heap.
esJavaOpts: "-Xmx128m -Xms128m"

# Allocate smaller chunks of memory per pod.
resources:
  requests:
    cpu: "100m"
    memory: "512M"
  limits:
    cpu: "1000m"
    memory: "512M"

# To allocate more space if above requests and limits aren't enough
# resources:
  # requests:
    # cpu: "500m"
    # memory: "2Gi"
  # limits:
    # cpu: "1000m"
    # memory: "5G"

# Request smaller persistent volumes.
volumeClaimTemplate:
  accessModes: [ "ReadWriteOnce" ]
  storageClassName: "efs-sc"
  resources:
    requests:
      storage: 100M

排查步骤

检查集群健康状态：执行以下命令查看集群初始化进度
```
kubectl exec -it elasticsearch-master-0 -- curl -u <elastic-user>:<elastic-password> http://localhost:9200/_cluster/health?pretty
```
重点关注status字段（red/yellow/green）和unassigned_shards数值，red状态说明主分片未分配，集群无法正常提供服务。
验证节点资源占用：用命令查看Pod的CPU和内存使用率
```
kubectl top pod elasticsearch-master-0
```
若内存使用率接近上限，说明资源配置不足导致节点无法完成初始化。
检查存储挂载情况：执行命令查看Pod的存储绑定详情
```
kubectl describe pod elasticsearch-master-0
```
确认efs-sc存储类对应的PVC是否成功绑定，挂载路径是否正常，同时注意EFS的访问模式是否适配多节点集群。
定位错误根源：日志中的GeoIP数据库获取失败是集群未就绪的结果，而非原因，核心问题是集群处于未恢复/未初始化状态。

解决建议

调整资源配置（核心修复）：Elasticsearch对内存要求较高，修改values-aws.yaml中的JVM参数和资源限制，满足最低运行要求：

esJavaOpts: "-Xmx2g -Xms2g" # JVM堆内存建议为物理内存的50%，且不超过32G
resources:
  requests:
    cpu: "500m"
    memory: "4Gi"
  limits:
    cpu: "1000m"
    memory: "4Gi"

修正存储访问模式：若部署多节点集群，EFS存储类需使用ReadWriteMany模式，否则仅单个节点能挂载存储，导致集群无法形成quorum：

volumeClaimTemplate:
  accessModes: [ "ReadWriteMany" ]
  storageClassName: "efs-sc"
  resources:
    requests:
      storage: 10Gi # 测试环境建议至少10G，避免空间不足

重新部署集群：先卸载现有实例，再重新安装

helm uninstall elasticsearch
helm install elasticsearch elastic/elasticsearch -f values-aws.yaml

等待集群就绪：执行命令监控StatefulSet的部署状态
```
kubectl rollout status statefulset/elasticsearch-master
```
待状态变为success后，再次执行集群健康检查确认状态为green。

可选：禁用GeoIP插件（若无需该功能）

plugins:
  install: [] # 不安装额外插件，包括GeoIP

内容的提问来源于stack exchange，提问作者Raj