redis-rdb-cli
A tool that can parse, filter, split, merge rdb and analyze memory usage offline. It can also sync 2 redis data and allow user define there own sink service to migrate redis data to somewhere.
![[粘贴]github-redis-rdb-cli_Redis](/i/li/?n=2&i=images/blog/202310/25121356_65389604920fa83918.svg?,size_16,text_QDUxQ1RP5Y2a5a6i,color_FFFFFF,t_30,g_se,x_10,y_10,shadow_20,type_ZmFuZ3poZW5naGVpdGk= "Java CI")
Chat with author
Contract the author
Binary release
Runtime requirement
jdk 1.8+
Install
$ wget https://github.com/leonchen83/redis-rdb-cli/releases/download/${version}/redis-rdb-cli-release.zip
$ unzip redis-rdb-cli-release.zip
$ cd ./redis-rdb-cli/bin
$ ./rct -h
Compile requirement
jdk 1.8+
maven-3.3.1+
Compile & run
$ git clone https://github.com/leonchen83/redis-rdb-cli.git
$ cd redis-rdb-cli
$ mvn clean install -Dmaven.test.skip=true
$ cd target/redis-rdb-cli-release/redis-rdb-cli/bin
$ ./rct -h
Run in docker
# run with jvm
$ docker run -it --rm redisrdbcli/redis-rdb-cli:latest
$ rct -V
# run without jvm
$ docker run -it --rm redisrdbcli/redis-rdb-cli:latest-native
$ rct -V
Build native image via graalvm in docker
$ docker build -m 8g -f DockerfileNative -t redisrdbcli:redis-rdb-cli .
$ docker run -it redisrdbcli:redis-rdb-cli bash
$ bash-5.1# rct -V
Windows Environment Variables
Add /path/to/redis-rdb-cli/bin to Path environment variable
Usage
Redis mass insertion
$ rct -f dump -s /path/to/dump.rdb -o /path/to/dump.aof -r
$ cat /path/to/dump.aof | /redis/src/redis-cli -p 6379 --pipe
Convert rdb to dump format
$ rct -f dump -s /path/to/dump.rdb -o /path/to/dump.aof
Convert rdb to json format
$ rct -f json -s /path/to/dump.rdb -o /path/to/dump.json
Numbers of key in rdb
$ rct -f count -s /path/to/dump.rdb -o /path/to/dump.csv
Find top 50 largest keys
$ rct -f mem -s /path/to/dump.rdb -o /path/to/dump.mem -l 50
Diff rdb
$ rct -f diff -s /path/to/dump1.rdb -o /path/to/dump1.diff
$ rct -f diff -s /path/to/dump2.rdb -o /path/to/dump2.diff
$ diff /path/to/dump1.diff /path/to/dump2.diff
Convert rdb to RESP
$ rct -f resp -s /path/to/dump.rdb -o /path/to/appendonly.aof
Sync with 2 redis
$ rst -s redis://127.0.0.1:6379 -m redis://127.0.0.1:6380 -r
Sync single redis to redis cluster
$ rst -s redis://127.0.0.1:6379 -m redis://127.0.0.1:30001 -r -d 0
Handle infinite loop in rst command
# set client-output-buffer-limit in source redis
$ redis-cli config set client-output-buffer-limit "slave 0 0 0"
$ rst -s redis://127.0.0.1:6379 -m redis://127.0.0.1:6380 -r
Migrate rdb to remote redis
$ rmt -s /path/to/dump.rdb -m redis://192.168.1.105:6379 -r
Downgrade migration
# Migrate data from redis-7 to redis-6
# About dump_rdb_version please see comment in redis-rdb-cli.conf
$ sed -i 's/dump_rdb_version=-1/dump_rdb_version=9/g' /path/to/redis-rdb-cli/conf/redis-rdb-cli.conf
$ rmt -s redis://com.redis7:6379 -m redis://com.redis6:6379 -r
Handle big key in migration
# set proto-max-bulk-len in target redis
$ redis-cli -h ${host} -p 6380 -a ${pwd} config set proto-max-bulk-len 2048mb
# set Xms Xmx in redis-rdb-cli node
$ export JAVA_TOOL_OPTIONS="-Xms8g -Xmx8g"
# execute migration
$ rmt -s redis://127.0.0.1:6379 -m redis://127.0.0.1:6380 -r
Migrate rdb to remote redis cluster
$ rmt -s /path/to/dump.rdb -c ./nodes-30001.conf -r
or simply use following cmd without nodes-30001.conf
$ rmt -s /path/to/dump.rdb -m redis://127.0.0.1:30001 -r
Backup remote rdb
$ rdt -b redis://192.168.1.105:6379 -o /path/to/dump.rdb
Backup remote rdb and convert db to dest db
$ rdt -b redis://192.168.1.105:6379 -o /path/to/dump.rdb --goal 3
Filter rdb
$ rdt -b /path/to/dump.rdb -o /path/to/filtered-dump.rdb -d 0 -t string
Split rdb via cluster's nodes.conf
$ rdt -s ./dump.rdb -c ./nodes.conf -o /path/to/folder -d 0
Merge multi rdb to one
$ rdt -m ./dump1.rdb ./dump2.rdb -o ./dump.rdb -t hash
Cut aof-use-rdb-preamble file to rdb file and aof file
$ rcut -s ./aof-use-rdb-preamble.aof -r ./dump.rdb -a ./appendonly.aof
Other parameter
More configurable parameter can be modified in /path/to/redis-rdb-cli/conf/redis-rdb-cli.conf
Filter
-
rct,rdtandrmtthese 3 commands support data filter bytype,dbandkeyRegEx(Java style). -
rstthis command only support data filter bydb.
For example:
$ rct -f dump -s /path/to/dump.rdb -o /path/to/dump.aof -d 0
$ rct -f dump -s /path/to/dump.rdb -o /path/to/dump.aof -t string hash
$ rmt -s /path/to/dump.rdb -m redis://192.168.1.105:6379 -r -d 0 1 -t list
$ rst -s redis://127.0.0.1:6379 -m redis://127.0.0.1:6380 -d 0
Monitor redis server
# step1
# open file `/path/to/redis-rdb-cli/conf/redis-rdb-cli.conf`
# change property `metric_gateway from `none` to `influxdb`
#
# step2
$ cd /path/to/redis-rdb-cli/dashboard
$ docker-compose up -d
#
# step3
$ rmonitor -s redis://127.0.0.1:6379 -n standalone
$ rmonitor -s redis://127.0.0.1:30001 -n cluster
$ rmonitor -s redis-sentinel://sntnl-usr:[email protected]:26379?master=mymaster&authUser=usr&authPassword=pwd -n sentinel
#
# step4
# open url `http://localhost:3000/d/monitor/monitor`, login grafana use `admin`, `admin` and check monitor result.
Difference between rmt and rst
- When
rmtstarted. source redis first doBGSAVEand generate a snapshot rdb file.rmtcommand migrate this snapshot file to target redis. after this process done,rmtterminated. -
rstnot only migrate snapshot rdb file but also incremental data from source redis. sorstnever terminated except typeCTRL+C.rstonly supportdbfilter more details please refer to Limitation of migration
Dashboard
Since v0.1.9, the rct -f mem support showing result in grafana dashboard like the following:
If you want to turn it on. you MUST install docker and docker-compose first, the installation please refer to dockerThen run the following command:
$ cd /path/to/redis-rdb-cli/dashboard
# start
$ docker-compose up -d
# stop
$ docker-compose down
cd /path/to/redis-rdb-cli/conf/redis-rdb-cli.confThen change parameter metric_gateway from none to influxdb.
Open http://localhost:3000 to check the rct -f mem's result.
If you deployed this tool in multi instance, you need to change parameter metric_instance to make sure unique between instances.
Redis 6
Redis 6 SSL
- use openssl to generate keystore
$ cd /path/to/redis-6.0-rc1
$ ./utils/gen-test-certs.sh
$ cd tests/tls
$ openssl pkcs12 -export -CAfile ca.crt -in redis.crt -inkey redis.key -out redis.p12
- If source redis and target redis use the same keystore. then config following parameters
source_keystore_path and target_keystore_path to point to/path/to/redis-6.0-rc1/tests/tls/redis.p12set source_keystore_pass and target_keystore_pass - after config ssl parameters use
rediss://host:portin your command to open ssl, for example:rst -s rediss://127.0.0.1:6379 -m rediss://127.0.0.1:30001 -r -d 0
Redis 6 ACL
- use following URI to open redis ACL support
$ rst -s redis://user:[email protected]:6379 -m redis://user:[email protected]:6380 -r -d 0
-
userMUST have+@allpermission to handle commands
Hack rmt
Rmt threading model
The rmt command use the following 4 parameters(redis-rdb-cli.conf) to migrate data to remote.
migrate_batch_size=4096
migrate_threads=4
migrate_flush=yes
migrate_retries=1
The most important parameter is migrate_threads=4. this means we use the following threading model to migrate data.
single redis ----> single redis
+--------------+ +----------+ thread 1 +--------------+
| | +----| Endpoint |-------------------| |
| | | +----------+ | |
| | | | |
| | | +----------+ thread 2 | |
| | |----| Endpoint |-------------------| |
| | | +----------+ | |
| Source Redis |----| | Target Redis |
| | | +----------+ thread 3 | |
| | |----| Endpoint |-------------------| |
| | | +----------+ | |
| | | | |
| | | +----------+ thread 4 | |
| | +----| Endpoint |-------------------| |
+--------------+ +----------+ +--------------+
single redis ----> redis cluster
+--------------+ +----------+ thread 1 +--------------+
| | +----| Endpoints|-------------------| |
| | | +----------+ | |
| | | | |
| | | +----------+ thread 2 | |
| | |----| Endpoints|-------------------| |
| | | +----------+ | |
| Source Redis |----| | Redis cluster|
| | | +----------+ thread 3 | |
| | |----| Endpoints|-------------------| |
| | | +----------+ | |
| | | | |
| | | +----------+ thread 4 | |
| | +----| Endpoints|-------------------| |
+--------------+ +----------+ +--------------+
The difference between cluster migration and single migration is Endpoint and Endpoints. In cluster migration the Endpoints contains multi Endpoint to point to every master instance in cluster. For example:
3 masters 3 replicas redis cluster. if migrate_threads=4 then we have 3 * 4 = 12 connections that connected with master instance.
Migration performance
The following 3 parameters affect migration performance
migrate_batch_size=4096
migrate_retries=1
migrate_flush=yes
-
migrate_batch_size: By default we use redispipelineto migrate data to remote. themigrate_batch_sizeis thepipelinebatch size. ifmigrate_batch_size=1then thepipelinedevolved into 1 single command to sent and wait the response from remote. -
migrate_retries: Themigrate_retries=1means if socket error occurred. we recreate a new socket and retry to send that failed command to target redis withmigrate_retriestimes. -
migrate_flush: Themigrate_flush=yesmeans we write every 1 command to socket. then we invokeSocketOutputStream.flush()immediately. ifmigrate_flush=nowe invokeSocketOutputStream.flush()when write to socket every 64KB. notice that this parameter also affectmigrate_retries. themigrate_retriesonly take effect whenmigrate_flush=yes.
Migration principle
+---------------+ +-------------------+ restore +---------------+
| | | redis dump format |---------------->| |
| | |-------------------| restore | |
| | convert | redis dump format |---------------->| |
| Dump rdb |------------>|-------------------| restore | Targe Redis |
| | | redis dump format |---------------->| |
| | |-------------------| restore | |
| | | redis dump format |---------------->| |
+---------------+ +-------------------+ +---------------+
Limitation of migration
- We use cluster's
nodes.confto migrate data to cluster. because of we didn't handle theMOVEDASKredirection. so limitation of cluster migration is that the cluster MUST in stable state during the migration. this means the cluster MUST have nomigrating,importingslot and no switch slave to master. - If use
rstmigrate data to cluster. the following commands not supportedPUBLISH,SWAPDB,MOVE,FLUSHALL,FLUSHDB,MULTI,EXEC,SCRIPT FLUSH,SCRIPT LOAD,EVAL,EVALSHA. and the following commandsRPOPLPUSH,SDIFFSTORE,SINTERSTORE,SMOVE,ZINTERSTORE,ZUNIONSTORE,DEL,UNLINK,RENAME,RENAMENX,PFMERGE,PFCOUNT,MSETNX,BRPOPLPUSH,BITOP,MSET,COPY,BLMOVE,LMOVE,ZDIFFSTORE,GEOSEARCHSTOREONLY SUPPORT WHEN THESE COMMAND KEYS IN THE SAME SLOT(eg:del {user}:1 {user}:2)
Hack ret
What ret command do
-
retcommand that allow user define there own sink service like sink redis data tomysqlormongodb. -
retcommand using Java SPI extension to do this job.
How to implement a sink service
User should follow the steps below to implement a sink service.
- create a java project using maven pom.xml
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0</modelVersion>
<groupId>com.your.company</groupId>
<artifactId>your-sink-service</artifactId>
<version>1.0.0</version>
<properties>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<maven.compiler.source>1.8</maven.compiler.source>
<maven.compiler.target>1.8</maven.compiler.target>
</properties>
<dependencies>
<dependency>
<groupId>com.moilioncircle</groupId>
<artifactId>redis-rdb-cli-api</artifactId>
<version>1.8.0</version>
<scope>provided</scope>
</dependency>
<dependency>
<groupId>com.moilioncircle</groupId>
<artifactId>redis-replicator</artifactId>
<version>[3.6.4, )</version>
<scope>provided</scope>
</dependency>
<dependency>
<groupId>org.slf4j</groupId>
<artifactId>slf4j-api</artifactId>
<version>1.7.25</version>
<scope>provided</scope>
</dependency>
<!--
<dependency>
other dependencies
</dependency>
-->
</dependencies>
<build>
<plugins>
<plugin>
<artifactId>maven-assembly-plugin</artifactId>
<version>3.1.0</version>
<configuration>
<descriptorRefs>
<descriptorRef>jar-with-dependencies</descriptorRef>
</descriptorRefs>
</configuration>
<executions>
<execution>
<id>make-assembly</id>
<phase>package</phase>
<goals>
<goal>single</goal>
</goals>
</execution>
</executions>
</plugin>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-compiler-plugin</artifactId>
<version>3.8.1</version>
<configuration>
<source>${maven.compiler.source}</source>
<target>${maven.compiler.target}</target>
<encoding>${project.build.sourceEncoding}</encoding>
</configuration>
</plugin>
</plugins>
</build>
</project>
- implement
SinkServiceinterface
public class YourSinkService implements SinkService {
@Override
public String sink() {
return "your-sink-service";
}
@Override
public void init(File config) throws IOException {
// parse your external sink config
}
@Override
public void onEvent(Replicator replicator, Event event) {
// your sink business
}
}
- register this service using Java SPI
# create com.moilioncircle.redis.rdb.cli.api.sink.SinkService file in src/main/resources/META-INF/services/
|-src
|____main
| |____resources
| | |____META-INF
| | | |____services
| | | | |____com.moilioncircle.redis.rdb.cli.api.sink.SinkService
# add following content in com.moilioncircle.redis.rdb.cli.api.sink.SinkService
your.package.YourSinkService
- package and deploy
$ mvn clean install
$ cp ./target/your-sink-service-1.0.0-jar-with-dependencies.jar /path/to/redis-rdb-cli/lib
- run your sink service
$ ret -s redis://127.0.0.1:6379 -c config.conf -n your-sink-service
- debug your sink service
public static void main(String[] args) throws Exception {
Replicator replicator = new RedisReplicator("redis://127.0.0.1:6379");
Runtime.getRuntime().addShutdownHook(new Thread(() -> {
Replicators.closeQuietly(replicator);
}));
replicator.addExceptionListener((rep, tx, e) -> {
throw new RuntimeException(tx.getMessage(), tx);
});
SinkService sink = new YourSinkService();
sink.init(new File("/path/to/your-sink.conf"));
replicator.addEventListener(new AsyncEventListener(sink, replicator, 4, Executors.defaultThreadFactory()));
replicator.open();
}
How to implement a formatter service
- create
YourFormatterServiceextendAbstractFormatterService
public class YourFormatterService extends AbstractFormatterService {
@Override
public String format() {
return "test";
}
@Override
public Event applyString(Replicator replicator, RedisInputStream in, int version, byte[] key, int type, ContextKeyValuePair context) throws IOException {
byte[] val = new DefaultRdbValueVisitor(replicator).applyString(in, version);
getEscaper().encode(key, getOutputStream());
getEscaper().encode(val, getOutputStream());
getOutputStream().write('\n');
return context;
}
}
- register this formatter using Java SPI
# create com.moilioncircle.redis.rdb.cli.api.format.FormatterService file in src/main/resources/META-INF/services/
|-src
|____main
| |____resources
| | |____META-INF
| | | |____services
| | | | |____com.moilioncircle.redis.rdb.cli.api.format.FormatterService
# add following content in com.moilioncircle.redis.rdb.cli.api.format.FormatterService
your.package.YourFormatterService
- package and deploy
$ mvn clean install
$ cp ./target/your-service-1.0.0-jar-with-dependencies.jar /path/to/redis-rdb-cli/lib
- run your formatter service
$ rct -f test -s redis://127.0.0.1:6379 -o ./out.csv -t string -d 0 -e json
Contributors
- Baoyi Chen
- Jintao Zhang
- Maz Ahmadi
- Anish Karandikar
- Air
- Raghu Nandan B S
- Special thanks to Kater Technologies
Consulting
Commercial support for redis-rdb-cli is available. The following services are currently available:
- Onsite consulting. $10,000 per day
- Onsite training. $10,000 per day
You may also contact Baoyi Chen directly, mail to [email protected].
Supported by 宁文君
27 January 2023, A sad day that I lost my mother 宁文君, She was encouraging and supporting me in developing this tool. Every time a company uses this tool, she got excited like a child and encouraged me to keep going. Without her I couldn't have maintained this tool for so many years. Even I didn't achieve much but she is still proud of me, R.I.P and hope God bless her.
Supported by IntelliJ IDEA
IntelliJ IDEA is a Java integrated development environment (IDE) for developing computer software.
It is developed by JetBrains (formerly known as IntelliJ), and is available as an Apache 2 Licensed community edition,
and in a proprietary commercial edition. Both can be used for commercial development.