redis平滑升级redis-2.4到redis-2.6

1. 升级的步骤和思路

    1.1 新建一个2.6.14版的redis。

    1.2 配置文件修改，作为slave进行配置。

    1.3 启动slave，并做好数据测试。

    1.4 将原来程序的redis读写迁移到新的slave上。

    1.5 升级原来的版本到2.6.14

2. redis 2.4和2.6的区别

    2.1 官方的2.4到2.6迁移说明

    2.2 redis2.6版的所有新特性和改进列表

升级前，系统环境：

192.168.1.119 , redis 2.4.15 , 单独的redis实例 

192.168.1.229 , 没有安装redis , 空闲的机器

升级中，系统环境：

192.168.1.119 , redis 2.4.15, redis master

192.168.1.229 , redis 2.6.14, redis slave

升级后，系统环境：

192.168.1.119 , redis 2.6.14, redis slave

192.168.1.229 , redis 2.6.14, redis master 

1. 升级的步骤和思路

1.1 新建一个2.6.14最新版redis

在192.168.1.229, 创建基本目录：

mkdir -p /opt/app/redis/{bin,db,logs,etc} 

安装

wget  http://redis.googlecode.com/files/redis-2.6.14.tar.gz

tar xzf redis-2.6.14.tar.gz

cd redis-2.6.14

make

cp src/redis-benchmark /opt/app/redis/bin/

cp src/redis-check-aof /opt/app/redis/bin/

cp src/redis-check-dump /opt/app/redis/bin/

cp src/redis-cli /opt/app/redis/bin/

cp src/redis-sentinel /opt/app/redis/bin/

cp src/redis-server /opt/app/redis/bin/

1.2 新建配置文件

/opt/app/redis/etc/redis.conf，内容如下：

daemonize yes

pidfile /var/run/redis.pid

port 6379

bind 192.168.1.229

timeout 300

loglevel warning

logfile /opt/app/redis/logs/redis.log

databases 16

save 900 1

save 300 10

save 60 10000

rdbcompression yes

dbfilename /opt/app/redis/db/redis.rdb

dir /opt/app/redis/db/

slaveof 192.168.1.119 6379

slave-serve-stale-data yes

maxmemory 128mb

appendonly no

appendfsync everysec

no-appendfsync-on-rewrite no

auto-aof-rewrite-percentage 100

auto-aof-rewrite-min-size 64mb

slowlog-log-slower-than 10000

slowlog-max-len 1024

hash-max-ziplist-entries 512

hash-max-ziplist-value 64

list-max-ziplist-entries 512

list-max-ziplist-value 64

set-max-intset-entries 512

zset-max-ziplist-entries 128

zset-max-ziplist-value 64

activerehashing yes

注意这三段红色部分：

第一段：表示redis绑定的ip和端口。

第二段：表示设置本redis为slave，并且设置redis的master为192.168.1.119 6379。redis 2.6的新特性，在slave的情况下，默认为只读，不能写入数据。我先以默认方式只读来同步数据，至于写入数据稍后在命令行里修改。

第三段：redis2.6新版中将选项名改为现在形式。

1.3 启动slave，数据测试

# 启动 slave

[root@localhost redis-2.6.14]# /opt/app/redis/bin/redis-server /opt/app/redis/etc/redis.conf

# 查看slave的info

[root@localhost redis-2.6.14]# /opt/app/redis/bin/redis-cli -h 192.168.1.229 info 

# Server
redis_version:2.6.14
redis_git_sha1:00000000
redis_git_dirty:0
redis_mode:standalone
os:Linux 2.6.18-238.el5 x86_64
arch_bits:64
multiplexing_api:epoll
gcc_version:4.1.2
process_id:14045
run_id:bd54d0a656843b62275888b1c22867856a61a50c
tcp_port:6379
uptime_in_seconds:76898
uptime_in_days:0
hz:10
lru_clock:905798

# Clients
connected_clients:2
client_longest_output_list:0
client_biggest_input_buf:0
blocked_clients:0

# Memory
used_memory:6347704
used_memory_human:6.05M
used_memory_rss:7655424
used_memory_peak:6367504
used_memory_peak_human:6.07M
used_memory_lua:31744
mem_fragmentation_ratio:1.21
mem_allocator:jemalloc-3.2.0

# Persistence
loading:0
rdb_changes_since_last_save:0
rdb_bgsave_in_progress:0
rdb_last_save_time:1372206578
rdb_last_bgsave_status:ok
rdb_last_bgsave_time_sec:0
rdb_current_bgsave_time_sec:-1
aof_enabled:0
aof_rewrite_in_progress:0
aof_rewrite_scheduled:0
aof_last_rewrite_time_sec:-1
aof_current_rewrite_time_sec:-1
aof_last_bgrewrite_status:ok

# Stats
total_connections_received:16
total_commands_processed:8238
instantaneous_ops_per_sec:0
rejected_connections:0
expired_keys:0
evicted_keys:0
keyspace_hits:66
keyspace_misses:8
pubsub_channels:0
pubsub_patterns:0
latest_fork_usec:6087

# Replication
role:slave
master_host:192.168.1.119
master_port:6379
master_link_status:up
master_last_io_seconds_ago:4
master_sync_in_progress:0
slave_priority:100
slave_read_only:1
connected_slaves:0

# CPU
used_cpu_sys:0.48
used_cpu_user:0.27
used_cpu_sys_children:0.61
used_cpu_user_children:0.50

# Keyspace
db0:keys=1328,expires=0
db1:keys=787,expires=0
db3:keys=123,expires=0
db4:keys=1,expires=0

当看到上面红色部分关于master信息时，表示slave已经将数据从master同步过来。

在reids2.6版中，slave默认是只读的。可以从命令行下可设置为同时读写。

[root@localhost redis-2.6.14]#  /opt/app/redis/bin/redis-cli -h 192.168.1.229 config set slave-read-only no

# 查看slave的info

[root@localhost redis-2.6.14]# /opt/app/redis/bin/redis-cli -h 192.168.1.229 info 

# Server
redis_version:2.6.14
redis_git_sha1:00000000
redis_git_dirty:0
redis_mode:standalone
os:Linux 2.6.18-238.el5 x86_64
arch_bits:64
multiplexing_api:epoll
gcc_version:4.1.2
process_id:14045
run_id:bd54d0a656843b62275888b1c22867856a61a50c
tcp_port:6379
uptime_in_seconds:76898
uptime_in_days:0
hz:10
lru_clock:905798

# Clients
connected_clients:2
client_longest_output_list:0
client_biggest_input_buf:0
blocked_clients:0

# Memory
used_memory:6347704
used_memory_human:6.05M
used_memory_rss:7655424
used_memory_peak:6367504
used_memory_peak_human:6.07M
used_memory_lua:31744
mem_fragmentation_ratio:1.21
mem_allocator:jemalloc-3.2.0

# Persistence
loading:0
rdb_changes_since_last_save:0
rdb_bgsave_in_progress:0
rdb_last_save_time:1372206578
rdb_last_bgsave_status:ok
rdb_last_bgsave_time_sec:0
rdb_current_bgsave_time_sec:-1
aof_enabled:0
aof_rewrite_in_progress:0
aof_rewrite_scheduled:0
aof_last_rewrite_time_sec:-1
aof_current_rewrite_time_sec:-1
aof_last_bgrewrite_status:ok

# Stats
total_connections_received:16
total_commands_processed:8238
instantaneous_ops_per_sec:0
rejected_connections:0
expired_keys:0
evicted_keys:0
keyspace_hits:66
keyspace_misses:8
pubsub_channels:0
pubsub_patterns:0
latest_fork_usec:6087

# Replication
role:slave
master_host:192.168.1.119
master_port:6379
master_link_status:up
master_last_io_seconds_ago:4
master_sync_in_progress:0
slave_priority:100
slave_read_only:0
connected_slaves:0

# CPU
used_cpu_sys:0.48
used_cpu_user:0.27
used_cpu_sys_children:0.61
used_cpu_user_children:0.50

# Keyspace
db0:keys=1328,expires=0
db1:keys=787,expires=0
db3:keys=123,expires=0
db4:keys=1,expires=0

1.4 更新应用程序的redis库

将php应用的redis连接修改为192.168.1.229上的redis

将Slave上的同步状态取消，避免主库在未完成版本升级前就重启，进而直接覆盖掉从库上的数据，导致所有的数据丢失。

redis 192.168.1.229:6379> SLAVEOF NO ONE

OK

1.5 升级原reids版本

升级192.168.1.119的redis版本到2.6.14

由于前面的操作，119上的redis可以放心的关闭和进行升级，操作和前面1-4步类似。

2. redis 2.4和2.6的区别

2.1 官方的2.4到2.6迁移说明：

Redis 2.4 is mostly a strict subset of 2.6. However there are a few things

that you should be aware of:

redis 2.4 很大部分上是 2.6 版的一个子集。然而还是有些需要注意的地方：

* You can't use .rdb and AOF files generated with 2.6 into a 2.4 instance.

* 2.6 slaves can be attached to 2.4 masters, but not the contrary, and only

  for the time needed to perform the version upgrade.

* 不可以把2.6生成的.rdb和AOF文件用于2.4的redis实例。

* 2.4版可以作为master让2.6版的slave连接，但是反过来不行。而且这样做仅仅在于进行版本升级的时候。

There are also a few API differences, that are unlikely to cause problems,

but it is better to keep them in mind:

还有一些API的小修改，这些不同不太可能造成什么问题，但是最好还是心里有数的好。

* SORT now will refuse to sort in numerical mode elements that can't be parsed  as numbers.

* EXPIREs now all have millisecond resolution (but this is very unlikely to

  break code that was not conceived exploting the previous resolution error

  in some way.)

* INFO output is a bit different now, and contains empty lines and comments

  starting with '#'. All the major clients should be already fixed to work

  with the new INFO format.

* Slaves are only read-only by default (but you can change this easily

  setting the "slave-read-only" configuration option to "no" editing your

  redis.conf or using CONFIG SET.

* SORT 将不对数字模式的元素进行排序，这些不能被解析为数字。

* EXPIRES 可以进行毫秒级的处理了。(这个和直接结束代码不同。直接结束代码在某些时候不会返回前一个处理的错误)

* INFO 的输出结果也有写改变。空行和注释以'#"开始。所有的主要clients(如php，python等)应该都已经和新的INFO格式兼容。

* "从"在默认情况下是只读的。(但是可以在配置文件中配置 slave-read-only no 或者在redis-cli命令行下使用config set命令配置为可写。)

The following INFO fields were renamed for consistency:

INFO格式中这些字段改名如下：

     changes_since_last_save -> rdb_changes_since_last_save

     bgsave_in_progress -> rdb_bgsave_in_progress

     last_save_time -> rdb_last_save_time

     last_bgsave_status -> rdb_last_bgsave_status

     bgrewriteaof_in_progress -> aof_rewrite_in_progress

     bgrewriteaof_scheduled -> aof_rewrite_scheduled

The following redis.conf and CONFIG GET / SET parameters changed:

redis.conf文件的配置项修改如下：

    * hash-max-zipmap-entries, now replaced by hash-max-ziplist-entries

    * hash-max-zipmap-value, now replaced by hash-max-ziplist-value

    * glueoutputbuf option was now completely removed (was deprecated)

2.2 An overview of new features and changes in Redis 2.6.x

======================================================

* Server side Lua scripting, see http://redis.io/commands/eval

* Virtual Memory removed (was deprecated in 2.4)

* Hardcoded limits about max number of clients removed.

* AOF low level semantics is generally more sane, and especially when used

  in slaves.

* Milliseconds resolution expires, also added new commands with milliseconds

  precision (PEXPIRE, PTTL, ...).

* Better memory usage for "small" lists, ziplists and hashes when fields or

  values contain small integers.

* Read only slaves.

* New bit opeations: BITCOUNT and BITOP commands.

* Clients max output buffer soft and hard limits. You can specifiy different

  limits for different classes of clients (normal,pubsub,slave).

* More incremental (less blocking) expired keys collection algorithm, in

  practical terms this means that Redis is more responsive when a very

  big number of keys expire about at the same time.

* AOF is now able to rewrite aggregate data types using variadic commands,

  often producing an AOF that is faster to save, load, and is smaller in size.

* Every redis.conf directive is now accepted as a command line option for the

  redis-server binary, with the same name and number of arguments.

* Hash table seed randomization for protection against collisions attacks.

* Performances improved when writing large objects to Redis.

* Integrated memory test, see redis-server --test-memory.

* INCRBYFLOAT and HINCRBYFLOAT commands.

* New DUMP, RESTORE, MIGRATE commands (back ported from Redis Cluster to 2.6).

* CRC64 checksump in RDB files.

* Better MONITOR output and behavior (now commands are logged before execution).

* "Software Watchdog" feature to debug latency issues.

* Significant parts of the core refactored or rewritten. New internal APIs

  and core changes allowed to develop Redis Cluster on top of the new code,

  however for 2.6 all the cluster code was removed, and will be released with

  Redis 3.0 when it is more complete and stable.

* Redis ASCII art logo added at startup.

* Crash report on memory violation or failed asserts improved significantly

  to make debugging of hard to catch bugs simpler.

* redis-benchmark improvements: ability to run selected tests,

  CSV output, faster, better help.

* redis-cli improvements: --eval for comfortable development of Lua scripts.

* SHUTDOWN now supports two optional arguments: "SAVE" and "NOSAVE".

* INFO output split into sections, the command is now able to just show 

  pecific sections.

* New statistics about how many time a command was called, and how much

  execution time it used (INFO commandstats).

* More predictable SORT behavior in edge cases.

* Better support for big endian and *BSD systems.

* Build system improved.