Redis Replication
Redis 의 FailOver 상황을 대비한 replication을 구성해 보겠습니다.
master 는 (read/write) 용이고, slave 는 master 의 데이터를 미러링하고 있는 read 전용입니다.
Redis Replication 아래와 같이 Master 1대, Slave 3대로 구성합니다.
Master : 10.0.0.201
Slave #1 : 10.0.0.211
Slave #2 : 10.0.0.212
Slave #3 : 10.0.0.213
Master 1대, Slave 3대에는 모두 Redis가 기본 설치 되어있어야 합니다.
Redis 기본 설치가 안되었으면 Redis 설치 클릭후 설치를 진행하시면 됩니다.
먼저 Master 구성 부터 진행하겠습니다.
1.Redis Replication Master Config
vi /etc/redis/6379.conf
- vi 로 아래와 같이 수정합니다.
bind 0.0.0.0 # redis Listen 대역을 0.0.0.0으로 수정 requirepass password # redis password masterauth password # requirepass 동일한 암호 repl-ping-slave-period 10 # 주석 해제 [마스터 서버와 동기화 주기] repl-timeout 60 # 주석 해제
- Redis 종료시 암호을 요구하기 때문에 redis start 스크립드에 암호 추가
vi /etc/init.d/redis_6379
- stop function 부분을 수정하시면 됩니다.
stop)
if [ ! -f $PIDFILE ]
then
echo "$PIDFILE does not exist, process is not running"
else
PID=$(cat $PIDFILE)
echo "Stopping ..."
$CLIEXEC -p $REDISPORT -a password shutdown # shutdown 시 암호입력
while [ -x /proc/${PID} ]
do
echo "Waiting for Redis to shutdown ..."
sleep 1
done
echo "Redis stopped"
fi
;;
2.Redis Replication Slave Config
vi /etc/redis/6379.conf
- vi 로 아래와 같이 수정합니다.
replicaof 10.0.0.201 6379 # master ip port bind 0.0.0.0 # redis Listen 대역을 0.0.0.0으로 수정 requirepass password # redis password masterauth password # requirepass 동일한 암호 repl-ping-slave-period 10 # 주석 해제 [마스터 서버와 동기화 주기] repl-timeout 60 # 주석 해제
- Redis 종료시 암호을 요구하기 때문에 redis start 스크립드에 암호 추가
stop)
if [ ! -f $PIDFILE ]
then
echo "$PIDFILE does not exist, process is not running"
else
PID=$(cat $PIDFILE)
echo "Stopping ..."
$CLIEXEC -p $REDISPORT -a password shutdown # shutdown 시 암호입력
while [ -x /proc/${PID} ]
do
echo "Waiting for Redis to shutdown ..."
sleep 1
done
echo "Redis stopped"
fi
;;
3.Redis Replication 확인
Redis Master [10.0.0.201] 을 시작합니다.
/etc/init.d/redis_6379 start
- tail 명령어로 로그를 확인합니다.
tail -50f /var/log/redis_6379.log
3004:M 18 Apr 2019 00:45:44.290 * Starting BGSAVE for SYNC with target: disk 3004:M 18 Apr 2019 00:45:44.290 * Background saving started by pid 3011 3011:C 18 Apr 2019 00:45:44.293 * DB saved on disk 3011:C 18 Apr 2019 00:45:44.294 * RDB: 0 MB of memory used by copy-on-write 3004:M 18 Apr 2019 00:45:44.320 * Background saving terminated with success 3004:M 18 Apr 2019 00:45:44.320 * Synchronization with replica 10.0.0.211:6379 succeeded 3004:M 18 Apr 2019 00:50:45.362 * Replica 10.0.0.212:6379 asks for synchronization 3004:M 18 Apr 2019 00:50:45.362 * Partial resynchronization request from 10.0.0.212:6379 accepted. Sending 420 bytes of backlog starting from offset 1. 3004:M 18 Apr 2019 00:52:02.065 * Replica 10.0.0.213:6379 asks for synchronization 3004:M 18 Apr 2019 00:52:02.065 * Partial resynchronization request from 10.0.0.213:6379 accepted. Sending 518 bytes of backlog starting from offset 1.
Redis Salve [10.0.0.211] 을 시작합니다.
/etc/init.d/redis_6379 start
- tail 명령어로 로그를 확인합니다.
tail -50f /var/log/redis_6379.log
2599:S 18 Apr 2019 00:45:44.032 # Server initialized 2599:S 18 Apr 2019 00:45:44.032 # WARNING you have Transparent Huge Pages (THP) support enabled in your kernel. This will create latency and memory usage issues with Redis. To fix this issue run the command 'echo never > /sys/kernel/mm/transparent_hugepage/enabled' as root, and add it to your /etc/rc.local in order to retain the setting after a reboot. Redis must be restarted after THP is disabled. 2599:S 18 Apr 2019 00:45:44.033 * DB loaded from disk: 0.001 seconds 2599:S 18 Apr 2019 00:45:44.033 * Before turning into a replica, using my master parameters to synthesize a cached master: I may be able to synchronize with the new master with just a partial transfer. 2599:S 18 Apr 2019 00:45:44.033 * Ready to accept connections 2599:S 18 Apr 2019 00:45:44.033 * Connecting to MASTER 10.0.0.201:6379 2599:S 18 Apr 2019 00:45:44.033 * MASTER <-> REPLICA sync started 2599:S 18 Apr 2019 00:45:44.034 * Non blocking connect for SYNC fired the event. 2599:S 18 Apr 2019 00:45:44.034 * Master replied to PING, replication can continue... 2599:S 18 Apr 2019 00:45:44.035 * Trying a partial resynchronization (request 2c8413e454e7a5e3ead2b9b66304e7aaa1541edd:1). 2599:S 18 Apr 2019 00:45:44.036 * Full resync from master: bf17ce558a05d7ecf614ebb8077b019134278479:0 2599:S 18 Apr 2019 00:45:44.036 * Discarding previously cached master state. 2599:S 18 Apr 2019 00:45:44.065 * MASTER <-> REPLICA sync: receiving 189 bytes from master 2599:S 18 Apr 2019 00:45:44.065 * MASTER <-> REPLICA sync: Flushing old data 2599:S 18 Apr 2019 00:45:44.065 * MASTER <-> REPLICA sync: Loading DB in memory 2599:S 18 Apr 2019 00:45:44.065 * MASTER <-> REPLICA sync: Finished with success
Redis Salve [10.0.0.212]을 시작합니다.
/etc/init.d/redis_6379 start
- tail 명령어로 로그를 확인합니다.
2598:S 18 Apr 2019 00:52:30.788 # Server initialized 2598:S 18 Apr 2019 00:52:30.788 # WARNING you have Transparent Huge Pages (THP) support enabled in your kernel. This will create latency and memory usage issues with Redis. To fix this issue run the command 'echo never > /sys/kernel/mm/transparent_hugepage/enabled' as root, and add it to your /etc/rc.local in order to retain the setting after a reboot. Redis must be restarted after THP is disabled. 2598:S 18 Apr 2019 00:52:30.789 * DB loaded from disk: 0.000 seconds 2598:S 18 Apr 2019 00:52:30.789 * Before turning into a replica, using my master parameters to synthesize a cached master: I may be able to synchronize with the new master with just a partial transfer. 2598:S 18 Apr 2019 00:52:30.789 * Ready to accept connections 2598:S 18 Apr 2019 00:52:30.789 * Connecting to MASTER 10.0.0.201:6379 2598:S 18 Apr 2019 00:52:30.789 * MASTER <-> REPLICA sync started 2598:S 18 Apr 2019 00:52:30.789 * Non blocking connect for SYNC fired the event. 2598:S 18 Apr 2019 00:52:30.790 * Master replied to PING, replication can continue... 2598:S 18 Apr 2019 00:52:30.791 * Trying a partial resynchronization (request bf17ce558a05d7ecf614ebb8077b019134278479:519). 2598:S 18 Apr 2019 00:52:30.791 * Successful partial resynchronization with master. 2598:S 18 Apr 2019 00:52:30.791 * MASTER <-> REPLICA sync: Master accepted a Partial Resynchronization.
Redis Salve [10.0.0.213]을 시작합니다.
/etc/init.d/redis_6379 start
- tail 명령어로 로그를 확인합니다.
2588:S 18 Apr 2019 00:53:17.777 # Server initialized 2588:S 18 Apr 2019 00:53:17.777 # WARNING you have Transparent Huge Pages (THP) support enabled in your kernel. This will create latency and memory usage issues with Redis. To fix this issue run the command 'echo never > /sys/kernel/mm/transparent_hugepage/enabled' as root, and add it to your /etc/rc.local in order to retain the setting after a reboot. Redis must be restarted after THP is disabled. 2588:S 18 Apr 2019 00:53:17.777 * DB loaded from disk: 0.001 seconds 2588:S 18 Apr 2019 00:53:17.777 * Before turning into a replica, using my master parameters to synthesize a cached master: I may be able to synchronize with the new master with just a partial transfer. 2588:S 18 Apr 2019 00:53:17.777 * Ready to accept connections 2588:S 18 Apr 2019 00:53:17.777 * Connecting to MASTER 10.0.0.201:6379 2588:S 18 Apr 2019 00:53:17.777 * MASTER <-> REPLICA sync started 2588:S 18 Apr 2019 00:53:17.778 * Non blocking connect for SYNC fired the event. 2588:S 18 Apr 2019 00:53:17.778 * Master replied to PING, replication can continue... 2588:S 18 Apr 2019 00:53:17.779 * Trying a partial resynchronization (request bf17ce558a05d7ecf614ebb8077b019134278479:589). 2588:S 18 Apr 2019 00:53:17.780 * Successful partial resynchronization with master. 2588:S 18 Apr 2019 00:53:17.780 * MASTER <-> REPLICA sync: Master accepted a Partial Resynchronization.
4.데이터를 확인
제대로 replication 이 되고 있다면, master 에 값을 넣었을 때 slave 들에도 해당 값이 들어가면 되겠습니다.
master 에 ‘foo’ 라는 key 로 값을 넣어 보겠습니다.
# Redis Master [10.0.0.201]
/usr/local/bin/redis-cli -a password Warning: Using a password with '-a' or '-u' option on the command line interface may not be safe. 127.0.0.1:6379> set foo var OK 127.0.0.1:6379> get foo "var" 127.0.0.1:6379>
# Redis Slave [10.0.0.211]
/usr/local/bin/redis-cli -a password Warning: Using a password with '-a' or '-u' option on the command line interface may not be safe. 127.0.0.1:6379> get foo "var" 127.0.0.1:6379>
slave 기본적으로 read 권한만 있습니다. 확인해보겠습니다.
127.0.0.1:6379> set foo var (error) READONLY You can't write against a read only replica. 127.0.0.1:6379>