从memcached命令行参数学起 - 1

查看memcached -h：

 

memcached 1.4.3 -p <num> TCP port number to listen on (default: 11211) -U <num> UDP port number to listen on (default: 11211, 0 is off) -s <file> UNIX socket path to listen on (disables network support) -a <mask> access mask for UNIX socket, in octal (default: 0700) -l <ip_addr> interface to listen on (default: INADDR_ANY, all addresses) -d run as a daemon -r maximize core file limit -u <username> assume identity of <username> (only when run as root) -m <num> max memory to use for items in megabytes (default: 64 MB) -M return error on memory exhausted (rather than removing items) -c <num> max simultaneous connections (default: 1024) -k lock down all paged memory. Note that there is a limit on how much memory you may lock. Trying to allocate more than that would fail, so be sure you set the limit correctly for the user you started the daemon with (not for -u <username> user; under sh this is done with 'ulimit -S -l NUM_KB'). -v verbose (print errors/warnings while in event loop) -vv very verbose (also print client commands/reponses) -vvv extremely verbose (also print internal state transitions) -h print this help and exit -i print memcached and libevent license -P <file> save PID in <file>, only used with -d option -f <factor> chunk size growth factor (default: 1.25) -n <bytes> minimum space allocated for key+value+flags (default: 48) -L Try to use large memory pages (if available). Increasing the memory page size could reduce the number of TLB misses and improve the performance. In order to get large pages from the OS, memcached will allocate the total item-cache in one large chunk. -D <char> Use <char> as the delimiter between key prefixes and IDs. This is used for per-prefix stats reporting. The default is ":" (colon). If this option is specified, stats collection is turned on automatically; if not, then it may be turned on by sending the "stats detail on" command to the server. -t <num> number of threads to use (default: 4) -R Maximum number of requests per event, limits the number of requests process for a given connection to prevent starvation (default: 20) -C Disable use of CAS -b Set the backlog queue limit (default: 1024) -B Binding protocol - one of ascii, binary, or auto (default) -I Override the size of each slab page. Adjusts max item size (default: 1mb, min: 1k, max: 128m) (END)

 

项目中使用到的启动参数：

/usr/local/memcached/bin/memcached -d -p 11911 -c 1024 -m 64

 

其中-d代表memcached作为daemon进程后台运行；-p指定监听的TCP端口，默认为11211；-c 指定最大软连接数目，默认为1024；-m 指定可用的最大内存，单位为M。

 

有以下问题：

1、-U 参数可以指定监听的UDP端口，tcp与udp可否同时使用？memcached针对这两种协议有哪些差别，会影响应用程序的哪些方面？

2、-c 为最大软连接数目，这个simultaneous connections与connections有什么差别？怎样定义？

3、-m 指定的最大内存，在什么情况下会达到？它与slab class、LRU之间的联系

 

为了查看memcached运行的情况，可以使用以下参数：

-v            verbose (print errors/warnings while in event loop)
-vv           very verbose (also print client commands/reponses)
-vvv          extremely verbose (also print internal state transitions)

 

1、查看UDP：

 

[flykobe@localhost ~]$ memcached -p 11211 -m 10 -c 10 -vvv slab class 1: chunk size 80 perslab 13107 slab class 2: chunk size 104 perslab 10082 slab class 3: chunk size 136 perslab 7710 slab class 4: chunk size 176 perslab 5957 slab class 5: chunk size 224 perslab 4681 slab class 6: chunk size 280 perslab 3744 slab class 7: chunk size 352 perslab 2978 。。。。。。 slab class 41: chunk size 717184 perslab 1 slab class 42: chunk size 1048576 perslab 1 <26 server listening (auto-negotiate) <27 server listening (auto-negotiate) <28 send buffer was 109568, now 268435456 <29 send buffer was 109568, now 268435456 <28 server listening (udp) <29 server listening (udp) <28 server listening (udp) <29 server listening (udp) <28 server listening (udp) <29 server listening (udp) <28 server listening (udp) <29 server listening (udp)

 

注意到最下面的“server listening (udp)”， 说明udp端口还是允许的。

 

[flykobe@localhost ~]$ memcached -p 11211 -m 10 -c 10 -vvv -U 0 slab class 1: chunk size 80 perslab 13107 slab class 2: chunk size 104 perslab 10082 slab class 3: chunk size 136 perslab 7710 slab class 4: chunk size 176 perslab 5957 slab class 5: chunk size 224 perslab 4681 slab class 6: chunk size 280 perslab 3744 slab class 7: chunk size 352 perslab 2978 slab class 8: chunk size 440 perslab 2383 slab class 9: chunk size 552 perslab 1899 。。。。。。 slab class 40: chunk size 573744 perslab 1 slab class 41: chunk size 717184 perslab 1 slab class 42: chunk size 1048576 perslab 1 <26 server listening (auto-negotiate) <27 server listening (auto-negotiate)

 

再运行：memcached -m 10 -c 10 -vvv -U 11311，指定memcached监听udp端口11311，并netstat：

 

[flykobe@localhost ~]$ netstat -a|grep 11311 tcp 0 0 *:11311 *:* LISTEN tcp 0 0 *:11311 *:* LISTEN udp 0 0 *:11311 *:* udp 0 0 *:11311 *:*

 

可以看到tcp与udp端口同时被监听。

 

为了进一步探究，先安装libmemcache：

wget　 "http://people.freebsd.org/~seanc/libmemcache/libmemcache-1.4.0.rc2.tar.bz2"

bzip2 -d libmemcache-1.4.0.rc2.tar.bz2

tar -xvf libmemcache-1.4.0.rc2.tar

./configure  --prefix=/usr/local/libmemcache --enable-debug

make

make install

 

在/usr/local下生成了libmemcache目录，其中包含Include和lib目录。

 

进入到libmemcache源码的test/benchmark目录，make benckmark 生成测试程序，这时可能需要配置下ld.so.conf，并且运行ldconfig。

 

[flykobe@localhost benchmark]$ ./benchmark 1 Value size: 50 Num tests: 1 Test Ops per second Total Time Time per Request set 3745.318352 0.000267 0.000267 get 2293.577982 0.000436 0.000436 add 6060.606061 0.000165 0.000165 delete 9803.921569 0.000102 0.000102

 

但是当./benchmark 的参数大于1，就会报错：

[FATAL@1261427683.457027] mcm_fetch_cmd():1176: memcache(4) protocol error: Operation now in progress: server sent data for key not in request
已放弃

 

修改代码：

if (strchr(tests, (int)'g') != NULL) { /* BEGIN get request */ if (gettimeofday(&t1, NULL) != 0) err(EX_OSERR, "gettimeofday(2)"); for (i = 0; i < num_tests; i++) { req = mc_req_new(); res = mc_req_add(req, key, keylen); res->size = valsize; res->val = malloc(res->size); mc_res_free_on_delete(res, 1); mc_get(mc, req); } if (gettimeofday(&t2, NULL) != 0) err(EX_OSERR, "gettimeofday(2)"); mc_req_free(req); /* END get test */ printf(fmt, "get", num_tests / tt(&t1, &t2), tt(&t1, &t2), tt(&t1, &t2) / num_tests); }

[flykobe@localhost benchmark]$ ./benchmark Value size: 50 Num tests: 10 Test Ops per second Total Time Time per Request set 12165.450122 0.000822 0.000082 get 9940.357853 0.001006 0.000101 add 4391.743522 0.002277 0.000228 delete 14727.540501 0.000679 0.000068

其中ops per second，指操作次数/s。

 

但是没能成功的用libmemcache，通过UDP协议，连接上memcached。

 

继续测试以下功能：

a、一个item字节的上限

b、存n个字节，需要的空间开销：size，使用的chunk

 

[flykobe@localhost benchmark]$ ./benchmark 10 1024 sg Value size: 1024 Num tests: 10 Test Ops per second Total Time Time per Request set 20920.502092 0.000478 0.000048 get 8389.261745 0.001192 0.000119 stats sizes STAT 1088 1 # 32字节(flag)+key长度+data长度 END stats slabs STAT 12:chunk_size 1096 STAT 12:chunks_per_page 956 STAT 12:total_pages 1 STAT 12:total_chunks 956 STAT 12:used_chunks 1 STAT 12:free_chunks 1 STAT 12:free_chunks_end 954 STAT 12:mem_requested 1085 STAT 12:get_hits 30 STAT 12:cmd_set 64 STAT 12:delete_hits 22 STAT 12:incr_hits 0 STAT 12:decr_hits 0 STAT 12:cas_hits 0 STAT 12:cas_badval 0 STAT active_slabs 2 STAT total_malloced 2096336 END

 

也就是说，为了存1024个字节的数据，真正使用到1088个字节，存储在最接近的slab，即1096字节的slab里。

 

从stats里可以看到，生成的最大slab块是1048576字节，使用上面一样的key，用下面的程序进行测试：

for(i = 1048001; i < valsize; i+=1){ val = (char *)malloc(i + 1); memset(val, 69, i); val[i] = '/0'; printf("i: %d/n", i); if ((ret=mc_set(mc, key, keylen, val, i, 0, 0)) < 0){ printf("quit: %d/n", ret); return -1; } free(val); }

 

运行结果为：最后打印的是 1 048 513 。由于1 048 513  + 32 + （1088 - 1024 - 32） = 1048577 字节，超过了1048576的上限，所以无法存储了！（1048576 = 2²⁰ = 1m） 也就是说，memcached存储单元的上限为1m-32-key的长度。