Memcache连接出错的问题
今天一早突然发现Memcache连接不到了，检查了一下Memcache的服务发现很正常。
root74090.00.01302961088?Ssl10:140:00memcached-d-m128-uroot-p11211-c1024-P/tmp/memcached.pid
root 7426 0.0 0.0 61220 764 pts/3 S+ 10:14 0:00 grep memcached既然Memcache很正常那么，为什么我的代码突然就连不上去了呢？
后来一想可能是谁把防火墙开启了，查看一下防火墙状态，出来了一大堆
表格：filter
ChainINPUT(policyACCEPT)
numtargetprotoptsourcedestination
1RH-Firewall-1-INPUTall--0.0.0.0/00.0.0.0/0
2ACCEPTtcp--0.0.0.0/00.0.0.0/0tcpdpt:3306
ChainFORWARD(policyACCEPT)
numtargetprotoptsourcedestination
1RH-Firewall-1-INPUTall--0.0.0.0/00.0.0.0/0
ChainOUTPUT(policyACCEPT)
numtargetprotoptsourcedestination
ChainRH-Firewall-1-INPUT(2references)
numtargetprotoptsourcedestination
1ACCEPTall--0.0.0.0/00.0.0.0/0
2ACCEPTicmp--0.0.0.0/00.0.0.0/0icmptype255
3ACCEPTesp--0.0.0.0/00.0.0.0/0
4ACCEPTah--0.0.0.0/00.0.0.0/0
5ACCEPTudp--0.0.0.0/0224.0.0.251udpdpt:5353
6ACCEPTudp--0.0.0.0/00.0.0.0/0udpdpt:631
7ACCEPTtcp--0.0.0.0/00.0.0.0/0tcpdpt:631
8ACCEPTall--0.0.0.0/00.0.0.0/0stateRELATED,ESTABLISHED
9ACCEPTtcp--0.0.0.0/00.0.0.0/0tcpdpt:3306
10ACCEPTtcp--0.0.0.0/00.0.0.0/0stateNEWtcpdpt:22
11ACCEPTtcp--0.0.0.0/00.0.0.0/0stateNEWtcpdpt:5900
12ACCEPTudp--0.0.0.0/00.0.0.0/0stateNEWudpdpt:5900
13 REJECT all -- 0.0.0.0/0 0.0.0.0/0 reject-with icmp-host-prohibited先把防火墙关掉试试结果可以连到。找到原因了，于是修改一下防火墙的规则，不让它拦截11211端口的连接就可以了。
[root@cloud1~]#/sbin/iptables-IINPUT-pudp--dport11211-jACCEPT
[root@cloud1~]#/etc/rc.d/init.d/iptablessave
将当前规则保存到/etc/sysconfig/iptables：[确定]
[root@cloud1~]#/etc/init.d/iptablesstatus
表格：filter
ChainINPUT(policyACCEPT)
numtargetprotoptsourcedestination
1ACCEPTudp--0.0.0.0/00.0.0.0/0udpdpt:11211
2ACCEPTtcp--0.0.0.0/00.0.0.0/0tcpdpt:11211
3RH-Firewall-1-INPUTall--0.0.0.0/00.0.0.0/0
4 ACCEPT tcp -- 0.0.0.0/0 0.0.0.0/0 tcp dpt:3306添加两条规则，然后保存一下就可以。
# /sbin/iptables -I INPUT -p tcp --dport 11211 -j ACCEPT
# /sbin/iptables -I INPUT -p udp --dport 11211 -j ACCEPT
# /etc/rc.d/init.d/iptables save
如果还不管用就重启一下防火墙
# /etc/init.d/iptables restart
memcahced的帮助
-p<num>TCPportnumbertolistenon(default:11211)
-U<num>UDPportnumbertolistenon(default:11211,0isoff)
-s<file>UNIXsocketpathtolistenon(disablesnetworksupport)
-a<mask>accessmaskforUNIXsocket,inoctal(default:0700)
-l<addr>interfacetolistenon(default:INADDR_ANY,alladdresses)
<addr>maybespecifiedashost:port.Ifyoudon'tspecify
aportnumber,thevalueyouspecifiedwith-por-Uis
used.Youmayspecifymultipleaddressesseparatedbycomma
orbyusing-lmultipletimes
-drunasadaemon
-rmaximizecorefilelimit
-u<username>assumeidentityof<username>(onlywhenrunasroot)
-m<num>maxmemorytouseforitemsinmegabytes(default:64MB)
-Mreturnerroronmemoryexhausted(ratherthanremovingitems)
-c<num>maxsimultaneousconnections(default:1024)
-klockdownallpagedmemory.Notethatthereisa
limitonhowmuchmemoryyoumaylock.Tryingto
allocatemorethanthatwouldfail,sobesureyou
setthelimitcorrectlyfortheuseryoustarted
thedaemonwith(notfor-u<username>user;
undershthisisdonewith'ulimit-S-lNUM_KB').
-vverbose(printerrors/warningswhileineventloop)
-vvveryverbose(alsoprintclientcommands/reponses)
-vvvextremelyverbose(alsoprintinternalstatetransitions)
-hprintthishelpandexit
-iprintmemcachedandlibeventlicense
-P<file>savePIDin<file>,onlyusedwith-doption
-f<factor>chunksizegrowthfactor(default:1.25)
-n<bytes>minimumspaceallocatedforkey+value+flags(default:48)
-LTrytouselargememorypages(ifavailable).Increasing
thememorypagesizecouldreducethenumberofTLBmisses
andimprovetheperformance.Inordertogetlargepages
fromtheOS,memcachedwillallocatethetotalitem-cache
inonelargechunk.
-D<char>Use<char>asthedelimiterbetweenkeyprefixesandIDs.
Thisisusedforper-prefixstatsreporting.Thedefaultis
":"(colon).Ifthisoptionisspecified,statscollection
isturnedonautomatically;ifnot,thenitmaybeturnedon
bysendingthe"statsdetailon"commandtotheserver.
-t<num>numberofthreadstouse(default:4)
-RMaximumnumberofrequestsperevent,limitsthenumberof
requestsprocessforagivenconnectiontoprevent
starvation(default:20)
-CDisableuseofCAS
-bSetthebacklogqueuelimit(default:1024)
-BBindingprotocol-oneofascii,binary,orauto(default)
-IOverridethesizeofeachslabpage.Adjustsmaxitemsize
(default:1mb,min:1k,max:128m)
-oCommaseparatedlistofextendedorexperimentaloptions
-(EXPERIMENTAL)maxconns_fast:immediatelyclosenew
connectionsifovermaxconnslimit
-hashpower:Anintegermultiplierforhowlargethehash
tableshouldbe.Canbegrownatruntimeifnotbigenough.
Setthisbasedon"STAThash_power_level"beforea
restart.memcached -d -m 256 -u root -l 10.1.106.20 -p 11211 -c 1024 -P /tmp/memcached.pid
-d选项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，这里是200MB
-u是运行Memcache的用户，如果当前为root的话，需要使用此参数指定用户。
-l是监听的服务器IP地址，如果有多个地址的话，我这里指定了服务器的IP地址192.168.1.91
-p是设置Memcache监听的端口，我这里设置了12301，最好是1024以上的端口
-c选项是最大运行的并发连接数，默认是1024，这里设置了256
-P是设置保存Memcache的pid文件，我这里是保存在/tmp/memcached.pid
停止Memcache进程：直接kill 进程号 或者 kill `cat /tmp/memcached.pid`查看启动的memcache服务：netstat -lp | grep memcached 或者 ps -ef | grep memcached