elasticsearch快照备份恢复

一、Repositories

在elasticsearch.yml文件中增加path.repo路径配置：
$ vim /etc/elasticsearch/elasticsearch.yml
path.repo: ["/home/ccx/es/backups", "/mount/longterm_backups"]

重启elasticsearch
$ service elasticsearch restart

创建文件夹作为repository的目录,并修改文件夹权限
$ mkdir -p /home/ccx/es/backups
$ chmod 755 /home/ccx/es/backups/
$ chown elasticsearch:elasticsearch /home/ccx/es/backups/

创建repository
$ curl -XPUT 'http://192.168.212.190:9200/_snapshot/my_backup' -d '{
"type": "fs",
"settings": {
"location": "/home/ccx/es/backups/my_backup",
"compress": true
}
}'
查看repository信息
$ curl -XGET 'http://192.168.212.190:9200/_snapshot/my_backup?pretty'

可以使用POST 请求，用来修改已经存在的repository
$ curl -XPOST 'http://192.168.212.190:9200/_snapshot/my_backup' -d '{
"type": "fs",
"settings": {
"location": "/home/ccx/es/backups/my_backup",
"max_snapshot_bytes_per_sec" : "50mb",
"max_restore_bytes_per_sec" : "50mb"

}
}'


二、Snapshot

创建快照
备份工作在后台运行
$ curl -XPUT 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_1'

同步执行,加wait_for_completion 标志,备份完成后才返回，如果数据量大的话，会花很长时间
$ curl -XPUT 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_2?wait_for_completion=true'

如果只想备份部分索引的话，可以加上indices 参数：
$ curl -XPUT 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_3' -d '{
"indices": "index_1,index_2",
"ignore_indices": "missing"
}'

查看备份信息
$ curl -XGET 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_2'

如果要查看所有索引的信息，使用如下api：
$ curl -XGET 'http://192.168.212.190:9200/_snapshot/my_backup/_all'

另外还有个一api可以看到更加详细的信息：
$ curl -XGET 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_2/_status'

删除备份
$ curl -XDELETE 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_2'


三、Restore

恢复snapshot_1里的全部索引：

$ curl -XPOST 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_1/_restore'

api额外的参数:

$ curl -XPOST 'http://192.168.212.190:9200/_snapshot/my_backup/snapshot_1/_restore' -d '{
"indices": "index_1",
"rename_pattern": "index_(.+)",
"rename_replacement": "restored_index_$1"
}'

参数indices 设置只恢复index_1索引，参数rename_pattern 和rename_replacement 用来正则匹配要恢复的索引，并且重命名。和备份一样，api会立刻返回值，然后在后台执行恢复，使用wait_for_completion 标记强制同步执行。

以使用下面两个api查看状态

$ curl -XGET 'http://192.168.212.190:9200/_recovery/'