SST(Search Sync Tool)

• 基于 medcl/esm 重构的 Search(Elasticsearch/OpenSearch) 同步工具

背景

• 因为要做集群灾备,需要在主备两个集群之间持续性地同步数据,调查过多个方案:
  • CCR(Cross-cluster replication) 官方工具,可惜需要收费,无奈放弃 :-(
  • esm 如官方文档说所,最大的特点快. 可惜也发现不少问题：
    • 同步后发现数据不全，一般需要采用多次执行的方式来补全数据;
    • 调查后发现，其采用的是获取源数据后，使用多个 goroutine 将数据通过bulk写入目标index, 会造成目标index重复写入很多相同数据的问题;
    • 更改源码,加入 --sync 功能，通过 scroll 同时查询源和目标index,比较其内容的方式,实现增量更新(Add/Update/Del), 目前该 PR/84 已经合并到esm中。
    • 目前 esm 的作者已经不再维护，因此后来发现的一些bug也很难改善及合并。
    • 从 issue 和 源码来看, esm 不支持 OpenSearch, esm作者以后也不会再更改

更改

• 对源码进行了更改,对同步(Sync)功能进行了加强,从而满足我在两个集群之间"近"实时同步的需求。
  • 增加 --stamp 参数，如果 index 中有表示最后更新时间的字段,可以进一步减少查询的数据量(本质是利用esm中的 --query 方式,单独提取出来更易于控制)
  • 增加对 OpenSearch 的支持
  • 对已有代码进行了较大的重构,更改了一些bug
  • 编写通用的集群同步脚本 search_sync.sh, 可以按需同步集群中的index,也可以对每月自动生成的 index 进行处理.
  • 为了方便测试,使用 docker-compose 来设置不同版本,不同类型的 源/目标 集群.

注意

• 目前主要的更改集中在 --sync 上,其他功能测试的不多,有可能会出现问题.

我是分割线(以下是原始文档,尚未更改)

---

Elasticsearch cross version data migration.

Links:

• Dec 3rd, 2020: [EN] Cross version Elasticsearch data migration with ESM
• Use INFINI Gateway to check the Document-Level differences between two clusters or indices after the migration

Features:

• Cross version migration supported
• Overwrite index name
• Copy index settings and mapping
• Support http basic auth
• Support dump index to local file
• Support loading index from local file
• Support http proxy
• Support sliced scroll ( elasticsearch 5.0 +)
• Support run in background
• Generate testing data by randomize the source document id
• Support rename filed name
• Support unify document type name
• Support specify which _source fields to return from source
• Support specify query string query to filter the data source
• Support rename source fields while do bulk indexing
• Support incremental update(add/update/delete changed records) with --sync. Notice: it use different implementation, just handle the changed records, but not as fast as the old way
• Load generating with

ESM is fast!

A 3 nodes cluster(3 * c5d.4xlarge， 16C，32GB，10Gbps)

root@ip-172-31-13-181:/tmp# ./esm -s https://localhost:8000 -d https://localhost:8000 -x logs1kw -y logs122 -m elastic:medcl123 -n elastic:medcl123 -w 40 --sliced_scroll_size=60 -b 5 --buffer_count=2000000 --regenerate_id
[12-19 06:31:20] [INF] [main.go:506,main] start data migration..
Scroll 10064570 / 10064570 [=================================================] 100.00% 55s
Bulk 10062602 / 10064570 [==================================================] 99.98% 55s
[12-19 06:32:15] [INF] [main.go:537,main] data migration finished.

Migrated 10,000,000 documents within a minute, Nginx log generated from kibana_sample_data_logs.

Before ESM

Before running the esm, please manually prepare the target index with mapping and optimized settings to improve the speed, for example:

PUT your-new-index
{
"settings": {
"index.translog.durability": "async", "refresh_interval": "-1", "number_of_shards": 10,
"number_of_replicas": 0
}
}

Example:

copy index index_name from 192.168.1.x to 192.168.1.y:9200

./bin/esm -s http://192.168.1.x:9200 -d http://192.168.1.y:9200 -x index_name -w=5 -b=10 -c 10000

copy index src_index from 192.168.1.x to 192.168.1.y:9200 and save with dest_index

./bin/esm -s http://localhost:9200 -d http://localhost:9200 -x src_index -y dest_index -w=5 -b=100

use sync feature for incremental update index src_index from 192.168.1.x to 192.168.1.y:9200

./bin/esm --sync -s http://localhost:9200 -d http://localhost:9200 -x src_index -y dest_index

support Basic-Auth

./bin/esm -s http://localhost:9200 -x "src_index" -y "dest_index" -d http://localhost:9201 -n admin:111111

copy settings and override shard size

./bin/esm -s http://localhost:9200 -x "src_index" -y "dest_index" -d http://localhost:9201 -m admin:111111 -c 10000 --shards=50 --copy_settings

copy settings and mapping, recreate target index, add query to source fetch, refresh after migration

./bin/esm -s http://localhost:9200 -x "src_index" -q=query:phone -y "dest_index" -d http://localhost:9201 -c 10000 --shards=5 --copy_settings --copy_mappings --force --refresh

dump elasticsearch documents into local file

./bin/esm -s http://localhost:9200 -x "src_index" -m admin:111111 -c 5000 -q=query:mixer --refresh -o=dump.bin 

dump source and target index to local file and compare them, so can find the difference quickly

./bin/esm --sort=_id -s http://localhost:9200 -x "src_index" --truncate_output --skip=_index -o=src.json
./bin/esm --sort=_id -s http://localhost:9200 -x "dst_index" --truncate_output --skip=_index -o=dst.json
diff -W 200 -ry --suppress-common-lines src.json dst.json

loading data from dump files, bulk insert to another es instance

./bin/esm -d http://localhost:9200 -y "dest_index" -n admin:111111 -c 5000 -b 5 --refresh -i=dump.bin

support proxy

 ./bin/esm -d http://123345.ap-northeast-1.aws.found.io:9200 -y "dest_index" -n admin:111111 -c 5000 -b 1 --refresh -i dump.bin --dest_proxy=http://127.0.0.1:9743

use sliced scroll(only available in elasticsearch v5) to speed scroll, and update shard number

 ./bin/esm -s=http://192.168.3.206:9200 -d=http://localhost:9200 -n=elastic:changeme -f --copy_settings --copy_mappings -x=bestbuykaggle --sliced_scroll_size=5 --shards=50 --refresh

migrate 5.x to 6.x and unify all the types to doc

./esm -s http://source_es:9200 -x "source_index*" -u "doc" -w 10 -b 10 - -t "10m" -d https://target_es:9200 -m elastic:passwd -n elastic:passwd -c 5000 

to migrate version 7.x and you may need to rename _type to _doc

./esm -s http://localhost:9201 -x "source" -y "target" -d https://localhost:9200 --rename="_type:type,age:myage" -u"_doc"

filter migration with range query

./esm -s https://192.168.3.98:9200 -m elastic:password -o json.out -x kibana_sample_data_ecommerce -q "order_date:[2020-02-01T21:59:02+00:00 TO 2020-03-01T21:59:02+00:00]"

range query, keyword type and escape

./esm -s https://192.168.3.98:9200 -m test:123 -o 1.txt -x test1 -q "@timestamp.keyword:[\"2021-01-17 03:41:20\" TO \"2021-03-17 03:41:20\"]"

generate testing data, if input.json contains 10 documents, the follow command will ingest 100 documents, good for testing

./bin/esm -i input.json -d http://localhost:9201 -y target-index1 --regenerate_id --repeat_times=10 

select source fields

 ./bin/esm -s http://localhost:9201 -x my_index -o dump.json --fields=author,title

rename fields while do bulk indexing

./bin/esm -i dump.json -d http://localhost:9201 -y target-index41 --rename=title:newtitle

user buffer_count to control memory used by ESM， and use gzip to compress network traffic

./esm -s https://localhost:8000 -d https://localhost:8000 -x logs1kw -y logs122 -m elastic:medcl123 -n elastic:medcl123 --regenerate_id -w 20 --sliced_scroll_size=60 -b 5 --buffer_count=1000000 --compress false 

Download

https://github.com/medcl/esm/releases

Compile:

if download version is not fill you environment,you may try to compile it yourself. go required.

make build

• go version >= 1.7

Options

Usage:
esm [OPTIONS]
Application Options:
-s, --source= source elasticsearch instance, ie: http://localhost:9200
-q, --query= query against source elasticsearch instance, filter data before migrate, ie: name:medcl
--sort= sort field when scroll, ie: _id (default: _id)
-d, --dest= destination elasticsearch instance, ie: http://localhost:9201
-m, --source_auth= basic auth of source elasticsearch instance, ie: user:pass
-n, --dest_auth= basic auth of target elasticsearch instance, ie: user:pass
-c, --count= number of documents at a time: ie "size" in the scroll request (10000)
--buffer_count= number of buffered documents in memory (100000)
-w, --workers= concurrency number for bulk workers (1)
-b, --bulk_size= bulk size in MB (5)
-t, --time= scroll time (1m)
--sliced_scroll_size= size of sliced scroll, to make it work, the size should be > 1 (1)
-f, --force delete destination index before copying
-a, --all copy indexes starting with . and _
--copy_settings copy index settings from source
--copy_mappings copy index mappings from source
--shards= set a number of shards on newly created indexes
-x, --src_indexes= indexes name to copy,support regex and comma separated list (_all)
-y, --dest_index= indexes name to save, allow only one indexname, original indexname will be used if not specified
-u, --type_override= override type name
--green wait for both hosts cluster status to be green before dump. otherwise yellow is okay
-v, --log= setting log level,options:trace,debug,info,warn,error (INFO)
-o, --output_file= output documents of source index into local file
--truncate_output= truncate before dump to output file
-i, --input_file= indexing from local dump file
--input_file_type= the data type of input file, options: dump, json_line, json_array, log_line (dump)
--source_proxy= set proxy to source http connections, ie: http://127.0.0.1:8080
--dest_proxy= set proxy to target http connections, ie: http://127.0.0.1:8080
--refresh refresh after migration finished
--sync= sync will use scroll for both source and target index, compare the data and sync(index/update/delete)
--fields= filter source fields(white list), comma separated, ie: col1,col2,col3,...
--skip= skip source fields(black list), comma separated, ie: col1,col2,col3,...
--rename= rename source fields, comma separated, ie: _type:type, name:myname
-l, --logstash_endpoint= target logstash tcp endpoint, ie: 127.0.0.1:5055
--secured_logstash_endpoint target logstash tcp endpoint was secured by TLS
--repeat_times= repeat the data from source N times to dest output, use align with parameter regenerate_id to amplify the data size
-r, --regenerate_id regenerate id for documents, this will override the exist document id in data source
--compress use gzip to compress traffic
-p, --sleep= sleep N seconds after finished a bulk request (-1)
Help Options:
-h, --help Show this help message

FAQ

• Scroll ID too long, update elasticsearch.yml on source cluster.

http.max_header_size: 16k
http.max_initial_line_length: 8k

Versions

From	To
1.x	1.x
1.x	2.x
1.x	5.x
1.x	6.x
1.x	7.x
2.x	1.x
2.x	2.x
2.x	5.x
2.x	6.x
2.x	7.x
5.x	1.x
5.x	2.x
5.x	5.x
5.x	6.x
5.x	7.x
6.x	1.x
6.x	2.x
6.x	5.0
6.x	6.x
6.x	7.x
7.x	1.x
7.x	2.x
7.x	5.x
7.x	6.x
7.x	7.x