1. Memcached简介

  • Memcached是一个开源的，高性能，分布式内存对象缓存系统。

  • Memcached是一种基于内存的key-value存储，用来存储小块的任意数据（字符串、对象）。这些数据可以是数据库调用、API调用或者是页面渲染的结果。

  • 一般的使用目的是，通过缓存数据库查询结果，减少数据库访问次数，以提高动态Web应用的速度、提高可扩展性。

1.1. 特征

memcached作为高速运行的分布式缓存服务器，具有以下的特点。

  • 协议简单
  • 基于libevent的事件处理
  • 内置内存存储方式
  • memcached不互相通信的分布式

2. 安装与运行

2.1. 自动安装

  1. # For Redhat/Fedora
  2. yum install -y memcached
  4. # For Debian or Ubuntu
  5. apt-get install memcached

2.2. 源码安装

安装指定版本的Memcached可以从 https://github.com/memcached/memcached/wiki/ReleaseNotes 地址下载。

  1. # Memcached depends on libevent
  2. yum install libevent-devel
  4. # install 
  5. wget https://memcached.org/latest
  6. [you might need to rename the file]
  7. tar -zxf memcached-1.x.x.tar.gz
  8. cd memcached-1.x.x
  9. ./configure --prefix=/usr/local/memcached
  10. make && make test && sudo make install

问题

如遇以下报错，可再执行make install。

  1. Signal handled: Interrupt.
  2. ok 51 - shutdown
  3. ok 52 - stop_server
  4. /bin/sh:行3: prove: 未找到命令
  5. make: *** [test] Error 127

2.3. 验证

确认是否安装成功，可执行以下命令

  1. /usr/local/memcached/bin/memcached -h

2.4. 运行

2.4.1. 前台运行

  1. /usr/local/memcached/bin/memcached -p 11211 -m 64m -vv

2.4.2. 后台运行

  1. /usr/local/memcached/bin/memcached -p 11211 -m 64m -d -c 102400 -t 8 -P /tmp/memcached.pid

2.5. 连接

  1. $ telnet 127.0.0.1 11211
  2. Trying 127.0.0.1...
  3. Connected to 127.0.0.1.
  4. Escape character is '^]'.
  5. set foo 0 0 3                                                   保存命令
  6. bar                                                             数据
  7. STORED                                                          结果
  8. get foo                                                         取得命令
  9. VALUE foo 0 3                                                   数据
  10. bar                                                             数据
  11. END                                                             结束行
  12. quit                                                            退出

3. Memcached运行参数

  1. # /usr/local/memcached/bin/memcached -h
  2. memcached 1.5.12
  3. -p, --port=<num>          TCP port to listen on (default: 11211)
  4. -U, --udp-port=<num>      UDP port to listen on (default: 0, off)
  5. -s, --unix-socket=<file>  UNIX socket to listen on (disables network support)
  6. -A, --enable-shutdown     enable ascii "shutdown" command
  7. -a, --unix-mask=<mask>    access mask for UNIX socket, in octal (default: 0700)
  8. -l, --listen=<addr>       interface to listen on (default: INADDR_ANY)
  9. -d, --daemon              run as a daemon
  10. -r, --enable-coredumps    maximize core file limit
  11. -u, --user=<user>         assume identity of <username> (only when run as root)
  12. -m, --memory-limit=<num>  item memory in megabytes (default: 64 MB)
  13. -M, --disable-evictions   return error on memory exhausted instead of evicting
  14. -c, --conn-limit=<num>    max simultaneous connections (default: 1024)
  15. -k, --lock-memory         lock down all paged memory
  16. -v, --verbose             verbose (print errors/warnings while in event loop)
  17. -vv                       very verbose (also print client commands/responses)
  18. -vvv                      extremely verbose (internal state transitions)
  19. -h, --help                print this help and exit
  20. -i, --license             print memcached and libevent license
  21. -V, --version             print version and exit
  22. -P, --pidfile=<file>      save PID in <file>, only used with -d option
  23. -f, --slab-growth-factor=<num> chunk size growth factor (default: 1.25)
  24. -n, --slab-min-size=<bytes> min space used for key+value+flags (default: 48)
  25. -L, --enable-largepages  try to use large memory pages (if available)
  26. -D <char>     Use <char> as the delimiter between key prefixes and IDs.
  27.               This is used for per-prefix stats reporting. The default is
  28.               ":" (colon). If this option is specified, stats collection
  29.               is turned on automatically; if not, then it may be turned on
  30.               by sending the "stats detail on" command to the server.
  31. -t, --threads=<num>       number of threads to use (default: 4)
  32. -R, --max-reqs-per-event  maximum number of requests per event, limits the
  33.                           requests processed per connection to prevent
  34.                           starvation (default: 20)
  35. -C, --disable-cas         disable use of CAS
  36. -b, --listen-backlog=<num> set the backlog queue limit (default: 1024)
  37. -B, --protocol=<name>     protocol - one of ascii, binary, or auto (default)
  38. -I, --max-item-size=<num> adjusts max item size
  39.                           (default: 1mb, min: 1k, max: 128m)
  40. -F, --disable-flush-all   disable flush_all command
  41. -X, --disable-dumping     disable stats cachedump and lru_crawler metadump
  42. -o, --extended            comma separated list of extended options
  43.                           most options have a 'no_' prefix to disable
  44.    - maxconns_fast:       immediately close new connections after limit
  45.    - hashpower:           an integer multiplier for how large the hash
  46.                           table should be. normally grows at runtime.
  47.                           set based on "STAT hash_power_level"
  48.    - tail_repair_time:    time in seconds for how long to wait before
  49.                           forcefully killing LRU tail item.
  50.                           disabled by default; very dangerous option.
  51.    - hash_algorithm:      the hash table algorithm
  52.                           default is murmur3 hash. options: jenkins, murmur3
  53.    - lru_crawler:         enable LRU Crawler background thread
  54.    - lru_crawler_sleep:   microseconds to sleep between items
  55.                           default is 100.
  56.    - lru_crawler_tocrawl: max items to crawl per slab per run
  57.                           default is 0 (unlimited)
  58.    - lru_maintainer:      enable new LRU system + background thread
  59.    - hot_lru_pct:         pct of slab memory to reserve for hot lru.
  60.                           (requires lru_maintainer)
  61.    - warm_lru_pct:        pct of slab memory to reserve for warm lru.
  62.                           (requires lru_maintainer)
  63.    - hot_max_factor:      items idle > cold lru age * drop from hot lru.
  64.    - warm_max_factor:     items idle > cold lru age * this drop from warm.
  65.    - temporary_ttl:       TTL's below get separate LRU, can't be evicted.
  66.                           (requires lru_maintainer)
  67.    - idle_timeout:        timeout for idle connections
  68.    - slab_chunk_max:      (EXPERIMENTAL) maximum slab size. use extreme care.
  69.    - watcher_logbuf_size: size in kilobytes of per-watcher write buffer.
  70.    - worker_logbuf_size:  size in kilobytes of per-worker-thread buffer
  71.                           read by background thread, then written to watchers.
  72.    - track_sizes:         enable dynamic reports for 'stats sizes' command.
  73.    - no_inline_ascii_resp: save up to 24 bytes per item.
  74.                            small perf hit in ASCII, no perf difference in
  75.                            binary protocol. speeds up all sets.
  76.    - no_hashexpand:       disables hash table expansion (dangerous)
  77.    - modern:              enables options which will be default in future.
  78.              currently: nothing
  79.    - no_modern:           uses defaults of previous major version (1.4.x)

常用参数：

  • -d是启动一个守护进程；
  • -m是分配给Memcache使用的内存数量，单位是MB；
  • -u是运行Memcache的用户；
  • -l是监听的服务器IP地址，可以有多个地址；
  • -p是设置Memcache监听的端口，，最好是1024以上的端口；
  • -c是最大运行的并发连接数，默认是1024；
  • -t是线程数，默认为4；
  • -P是设置保存Memcache的pid文件。

参考文章：