第1.5 章 elasticsearch备份过程记录

warrah

已于 2024-10-29 10:54:14 修改

阅读量9.3k

点赞数 4

分类专栏：岁月云——大数据杂烩文章标签： elasticsearch

于 2018-01-05 10:30:18 首次发布

本文链接：https://blog.csdn.net/warrah/article/details/78978244

版权

岁月云——大数据杂烩专栏收录该内容

72 篇文章 3 订阅

订阅专栏

每个人使用elasticsearch的起点不一样，遇到的问题也参差不齐。
一开始elasticsearch不是我安装，配置信息如下:

cluster.name: dzm_dev
node.name: node-185
network.host: 192.168.5.185
http.port: 9200
transport.tcp.port: 9300
http.cors.enabled: true
http.cors.allow-origin: "*"
node.master: true
node.data: true
discovery.zen.ping.unicast.hosts: ["192.168.5.185","192.168.5.186", "192.168.5.187"]

如果我想备份，那么我先需要在elasticsearch.yml中添加一行配置path.repo: [“/usr/hadoop/application/el_bak”]这个’/usr/hadoop/application/el_bak’就是备份文件的路径，根据自己需要指定，不过一定要注意用户权限。在权限内，这个文件夹会自动创建。
配置好后，先删掉elasticsearch进程，然后再启动elasticsearch。

/usr/hadoop/application/elasticsearch/bin/elasticsearch -d
#日志
tail -fn 100 /usr/hadoop/application/elasticsearch/logs/dzm_dev.log

还有一点需要注意，因为discovery.zen.ping.unicast.hosts: ["192.168.5.185","192.168.5.186", "192.168.5.187"]这里面的其他机器也都需要加上path.repo的配置，否则创建仓库的时候，会提示下面的错误

org.elasticsearch.transport.RemoteTransportException: [node-187][192.168.5.187:9300][cluster:admin/repository/put]
Caused by: org.elasticsearch.repositories.RepositoryException: [el_back] failed to create repository
    at org.elasticsearch.repositories.RepositoriesService.createRepository(RepositoriesService.java:388) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.repositories.RepositoriesService.registerRepository(RepositoriesService.java:356) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.repositories.RepositoriesService.access$100(RepositoriesService.java:56) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.repositories.RepositoriesService$1.execute(RepositoriesService.java:109) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.ClusterStateUpdateTask.execute(ClusterStateUpdateTask.java:45) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.service.ClusterService.executeTasks(ClusterService.java:634) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.service.ClusterService.calculateTaskOutputs(ClusterService.java:612) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.service.ClusterService.runTasks(ClusterService.java:571) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.service.ClusterService$ClusterServiceTaskBatcher.run(ClusterService.java:263) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.service.TaskBatcher.runIfNotProcessed(TaskBatcher.java:150) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.cluster.service.TaskBatcher$BatchedTask.run(TaskBatcher.java:188) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.common.util.concurrent.ThreadContext$ContextPreservingRunnable.run(ThreadContext.java:569) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.runAndClean(PrioritizedEsThreadPoolExecutor.java:247) ~[elasticsearch-5.5.0.jar:5.5.0]
    at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.run(PrioritizedEsThreadPoolExecutor.java:210) ~[elasticsearch-5.5.0.jar:5.5.0]

当我把185,186,187三台机器都配置好后，创建操作，结果提示下面的错误

[2018-01-05T10:24:34,860][WARN ][o.e.r.VerifyNodeRepositoryAction] [node-185] [el_back] failed to verify repository
org.elasticsearch.repositories.RepositoryVerificationException: [el_back] a file written by master to the store [/usr/hadoop/application/el_bak] cannot be accessed on the node [{node-185}{yyAyGv1mQcKEsGDVMa84ZQ}{E5jcL1_5R3CBSzhrwJQ8mQ}{192.168.5.185}{192.168.5.185:9300}]. This might indicate that the store [/usr/hadoop/application/el_bak] is not shared between this node and the master node or that permissions on the store don't allow reading files written by the master node
	at org.elasticsearch.repositories.blobstore.BlobStoreRepository.verify(BlobStoreRepository.java:1025) ~[elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.repositories.VerifyNodeRepositoryAction.doVerify(VerifyNodeRepositoryAction.java:117) ~[elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.repositories.VerifyNodeRepositoryAction.access$300(VerifyNodeRepositoryAction.java:50) ~[elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.repositories.VerifyNodeRepositoryAction$VerifyNodeRepositoryRequestHandler.messageReceived(VerifyNodeRepositoryAction.java:153) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.repositories.VerifyNodeRepositoryAction$VerifyNodeRepositoryRequestHandler.messageReceived(VerifyNodeRepositoryAction.java:148) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.transport.TransportRequestHandler.messageReceived(TransportRequestHandler.java:33) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.transport.RequestHandlerRegistry.processMessageReceived(RequestHandlerRegistry.java:69) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.transport.TcpTransport$RequestHandler.doRun(TcpTransport.java:1544) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.common.util.concurrent.EsExecutors$1.execute(EsExecutors.java:110) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.transport.TcpTransport.handleRequest(TcpTransport.java:1501) [elasticsearch-5.5.0.jar:5.5.0]
	at org.elasticsearch.transport.TcpTransport.messageReceived(TcpTransport.java:1385) [elasticsearch-5.5.0.jar:5.5.0]

在elasticsearch5使用snapshot接口备份索引我找到了答案，我的elasticsearch集群需要配置共享.
1 采用nfs挂载
参考Linux—centos安装配置并挂载NFS
注意vim /etc/exports，内容如下,其中/usr/hadoop/application/el_bak是挂载的目录,括号里面是对权限进行设置。

/usr/hadoop/application/el_bak *(rw,sync,no_root_squash)

下面是centos6下面的操作，
如果没有安装nfs，可以通过yum -y install nfs-utils rpcbind,参考
Linux—centos安装配置并挂载NFS

185服务端
/etc/init.d/nfs status
/etc/init.d/rpcbind status
# 未启动服务的要先启动服务
/etc/init.d/nfs start
/etc/init.d/rpcbind start
#编辑配置文件
vi /etc/exports
/usr/hadoop/application/el_bak *(rw,sync,no_root_squash)
186客户端
/etc/init.d/rpcbind status
# 未启动服务的要先启动服务
/etc/init.d/rpcbind start
#挂载
showmount -e 192.168.5.185 
mount -t nfs 192.168.5.185 
:/usr/hadoop/application/el_bak /usr/hadoop/application/el_bak
#卸载
umount /usr/hadoop/application/el_bak

在centos7下面nfs有所区别，通过yum install -y nfs-utils进行安装，参考Centos7安装配置NFS服务和挂载

#在master节点上执行
#先为rpcbind和nfs做开机启动：(必须先启动rpcbind服务)
systemctl enable rpcbind.service
systemctl enable nfs-server.service
#然后分别启动rpcbind和nfs服务：
systemctl start rpcbind.service
systemctl start nfs-server.service
#确认NFS服务器启动成功：
rpcinfo -p
# 使配置生效
exportfs -r
# 查看配置情况
exportfs

#在slave节点，不需要启动nfs服务
 systemctl enable rpcbind.service
 systemctl start rpcbind.service
 showmount -e 192.168.5.185 
 mount -t nfs 192.168.5.185 :/application/el_bak /application/el_bak

2 采用sshfs挂载
参考ElasticSearch集群数据迁移备份方案
3 快照仓库操作REST接口
删除仓库

DELETE http://192.168.5.185:9200/_snapshot/my_backup

创建仓库

curl -XPUT 'http://192.168.5.185:9200/_snapshot/el_bak' -d '{"type": "fs","settings": {"location":"/usr/hadoop/application/el_bak","max_snapshot_bytes_per_sec" : "50mb", "max_restore_bytes_per_sec" :"50mb","compress":true}}'

创建的时候出现了下面的错误

{"error":{"root_cause":[{"type":"repository_verification_exception","reason":"[el_bak] [[yyAyGv1mQcKEsGDVMa84ZQ, 'RemoteTransportException[[node-185][192.168.5.185:9300][internal:admin/repository/verify]]; nested: RepositoryVerificationException[[el_bak] store location [/usr/hadoop/application/el_bak] is not accessible on the node [{node-185}{yyAyGv1mQcKEsGDVMa84ZQ}{1yS2vGSXQjSRYjraptiIcw}{192.168.5.185}{192.168.5.185:9300}]]; nested: AccessDeniedException[/usr/hadoop/application/el_bak/tests-km6E2_MpQ2aE4REyClbPTg/data-yyAyGv1mQcKEsGDVMa84ZQ.dat];'], [QjyiZHI8R1SnJfue-7x6Bg, 'RemoteTransportException[[node-187][192.168.5.187:9300][internal:admin/repository/verify]]; nested: RepositoryVerificationException[[el_bak] store location [/usr/hadoop/application/el_bak] is not accessible on the node [{node-187}{QjyiZHI8R1SnJfue-7x6Bg}{cLXcXL7XR16WokqHnwZ14g}{192.168.5.187}{192.168.5.187:9300}]]; nested: AccessDeniedException[/usr/hadoop/application/el_bak/tests-km6E2_MpQ2aE4REyClbPTg/data-QjyiZHI8R1SnJfue-7x6Bg.dat];']]"}],"type":"repository_verification_exception","reason":"[el_bak] [[yyAyGv1mQcKEsGDVMa84ZQ, 'RemoteTransportException[[node-185][192.168.5.185:9300][internal:admin/repository/verify]]; nested: RepositoryVerificationException[[el_bak] store location [/usr/hadoop/application/el_bak] is not accessible on the node [{node-185}{yyAyGv1mQcKEsGDVMa84ZQ}{1yS2vGSXQjSRYjraptiIcw}{192.168.5.185}{192.168.5.185:9300}]]; nested: AccessDeniedException[/usr/hadoop/application/el_bak/tests-km6E2_MpQ2aE4REyClbPTg/data-yyAyGv1mQcKEsGDVMa84ZQ.dat];'], [QjyiZHI8R1SnJfue-7x6Bg, 'RemoteTransportException[[node-187][192.168.5.187:9300][internal:admin/repository/verify]]; nested: RepositoryVerificationException[[el_bak] store location [/usr/hadoop/application/el_bak] is not accessible on the node [{node-187}{QjyiZHI8R1SnJfue-7x6Bg}{cLXcXL7XR16WokqHnwZ14g}{192.168.5.187}{192.168.5.187:9300}]]; nested: AccessDeniedException[/usr/hadoop/application/el_bak/tests-km6E2_MpQ2aE4REyClbPTg/data-QjyiZHI8R1SnJfue-7x6Bg.dat];']]"},"status":500}

但是有可以查询到，很是奇怪了

跟踪el_bak文件的权限，可以看到el_bak所属用户居然会自己变，很有意思。

通过cat /etc/passwd命令查看到三台机器的hadoop用户名对应的id不一致，所以造成了上面的情况。

这个问题比较隐蔽，先将对用户下面的进程全部关闭掉，然后更改\etc\password中的pid为相同的，重启elasticsearch在执行就ok了。

 curl -XPUT 'http://192.168.5.185:9200/_snapshot/el_bak' -d '{"type": "fs","settings": {"location":"/usr/hadoop/application/el_bak","max_snapshot_bytes_per_sec" : "50mb", "max_restore_bytes_per_sec" :"50mb","compress":true}}'

4 针对索引创建备份

5 备份还原

POST http://192.168.5.185:9200/_snapshot/el_bak/20180109_3/_restore

6 删除type，然后想还原索引带来的问题

这个时候如果执行版本恢复，则提示下面的错误

{
  "error": {
    "root_cause": [
      {
        "type": "snapshot_restore_exception",
        "reason": "[el_bak:20180109_3/W8DWSBr1TXOX3tDwUocTjQ] cannot restore index [xxxinfo] because it's open"
      }
    ],
    "type": "snapshot_restore_exception",
    "reason": "[el_bak:20180109_3/W8DWSBr1TXOX3tDwUocTjQ] cannot restore index [xxxinfo] because it's open"
  },
  "status": 500
}

看来需要删掉索引，然后通过索引恢复才是正确的
我在生产环境中执行，在备份时还出现这样的问题，不得不再搜索资料。那是因为仓库没有创建

{
  "error": {
    "root_cause": [
      {
        "type": "repository_missing_exception",
        "reason": "[el_bak] missing"
      }
    ],
    "type": "repository_missing_exception",
    "reason": "[el_bak] missing"
  },
  "status": 404
}

7 elasticdump
elasticsearch迁移工具–elasticdump的使用

nohup elasticdump \
  --input=http://elastic:test@172.19.1.153:9200/test \
  --searchBody='{
    "query": {
      "range": {
        "createdate": {
          "lte": "2020-07-20 23:59:59"
        }
      }
    }
  }' \
  --output=/appdata/test_bak_20200720.json \
  --limit=2000 \
  --type=data &

如果只想获取部分字段，则参考

nohup elasticdump --input=http://elastic:password@172.19.11.153:9200/test  --output=test.json --sourceOnly --searchBody='
{
  "query": {
    "bool": {
      "must": [
        {
          "match_phrase": 
          {"xhdwsbh": "11111"}
          
        },
        {
          "range": {
            "kprq": {
              "gte": "20191105",
              "lte": "20191106"
            }
          }
        }
      ]
    }
    
    
  } ,
  "_source": ["fplxdm","fpdm","fphm","fpzt","kprq","xhdwsbh","xhdwmc","xhdwdzdh","xhdwyhzh","ghdwsbh","ghdwmc","ghdwdzdh","ghdwyhzh","hsslbs","zhsl","hjje","hjse","jshj","bz","zyspmc"]
}' &

直接从一台机器到另外一台机器，要求版本一直

elasticdump \
--input=http://elastic:密码@10.101.10.58:9200/5nfo \
--output=http://elastic:密码@10.101.10.55:9200/5info \
--type=data --searchBody '{"query":{"match_all": {}}}'

压缩,elasticdump速度还有有些慢

yum install nodejs -y
yum install npm --enablerepo=epel -y
npm install -g cnpm --registry=https://registry.npm.taobao.org

export PATH=$PATH:/root/node_modules/elasticdump/bin
source /etc/profile


nohup elasticdump \
  --input=http://192.168.0.180:9200/invoice_title \
  --type=data \
  --limit=1000 \
  --output=$ \
  | gzip > /appdata/nfsv/elastic/invoice_title_202201201.json.gz &

分片备份脚本，在源es集群中执行

#!/bin/bash

# 设置环境变量
export ELASTICSEARCH_USERNAME=es用户
export ELASTICSEARCH_PASSWORD=es密码

SOURCE_URL="http://10.101.10.55:9201"
TARGET_URL="http://10.101.10.2:9200"
INDEX_NAME="invoice_title_v3"
BATCH_SIZE=10000
FROM=0

# 获取总文档数
TOTAL_COUNT=$(curl -s -u $ELASTICSEARCH_USERNAME:$ELASTICSEARCH_PASSWORD -X POST "$SOURCE_URL/$INDEX_NAME/_count" -H 'Content-Type: application/json' -d '{"query":{"match_all": {}}}' | jq -r '.count')

if [ "$TOTAL_COUNT" -eq 0 ]; then
  echo "No documents to migrate. Exiting."
  exit 0
fi

echo "Total documents to migrate: $TOTAL_COUNT"

# 创建日志文件
LOG_FILE="migration_log_$(date +%Y%m%d_%H%M%S).log"
echo "Migration started at $(date)" > $LOG_FILE

while [ $FROM -lt $TOTAL_COUNT ]; do
  echo "Migrating batch from $FROM" | tee -a $LOG_FILE
  elasticdump \
    --input=http://es用户:es密码@10.101.10.55:9201/invoice_title_v3 \
    --output=http://es用户:es密码@10.101.10.2:9200/invoice_title_v3 \
    --type=data \
    --searchBody "{\"query\":{\"match_all\": {}}, \"size\": $BATCH_SIZE, \"from\": $FROM}" \
    --limit $BATCH_SIZE \
    --ignore-errors 

  if [ $? -ne 0 ]; then
    echo "Error migrating batch from $FROM" | tee -a $LOG_FILE
    exit 1
  fi

  FROM=$((FROM + BATCH_SIZE))
done

echo "Data migration completed successfully." | tee -a $LOG_FILE
echo "Migration ended at $(date)" >> $LOG_FILE