Haproxy-调度算法详解

HAProxy调度算法

HAProxy通过固定参数 balance 指明对后端服务器的调度算法,该参数可以配置在listen或backend选项中。

HAProxy的调度算法分为静态和动态调度算法,但是有些算法可以根据参数在静态和动态算法中相互转换。

官方文档:http://cbonte.github.io/haproxy-dconv/2.1/configuration.html#4-balance

静态算法

静态算法:按照事先定义好的规则轮询公平调度,不关心后端服务器的当前负载、链接数和响应速度等,且无法实时修改权重,只能靠重启HAProxy生效。

可以利用 socat工具对服务器动态权重和其它状态的调整,Socat 是 Linux 下的一个多功能的网络工具,名字来由是Socket CAT,Socat 的主要特点就是在两个数据流之间建立通道,且支持众多协议和链接方式。如 IP、TCP、 UDP、IPv6、Socket文件等

范例:利用工具socat 对服务器动态权重调整

[root@centos7 ~]#yum -y install socat
#查看帮助
[root@centos7 ~]#socat -h
[root@centos7 ~]#echo "help" | socat stdio /var/lib/haproxy/haproxy.sock
Unknown command. Please enter one of the following commands only :
help           : this message
prompt         : toggle interactive mode with prompt
quit           : disconnect
show tls-keys [id|*]: show tls keys references or dump tls ticket keys when id specified
set ssl tls-key [id|keyfile] <tlskey>: set the next TLS key for the <id> or <keyfile> listener to <tlskey>
set ssl cert <certfile> <payload> : replace a certificate file
commit ssl cert <certfile> : commit a certificate file
abort ssl cert <certfile> : abort a transaction for a certificate file
show sess [id] : report the list of current sessions or dump this session
shutdown session : kill a specific session
shutdown sessions server : kill sessions on a server
clear counters : clear max statistics counters (add 'all' for all counters)
show info      : report information about the running process [desc|json|typed]*
show stat      : report counters for each proxy and server [desc|json|typed]*
show schema json : report schema used for stats
disable agent  : disable agent checks (use 'set server' instead)
disable health : disable health checks (use 'set server' instead)
disable server : disable a server for maintenance (use 'set server' instead)
enable agent   : enable agent checks (use 'set server' instead)
enable health  : enable health checks (use 'set server' instead)
enable server  : enable a disabled server (use 'set server' instead)
set maxconn server : change a server's maxconn setting
set server     : change a server's state, weight or address
get weight     : report a server's current weight
set weight     : change a server's weight (deprecated)
show startup-logs : report logs emitted during HAProxy startup
show peers [peers section]: dump some information about all the peers or this peers section
set maxconn global : change the per-process maxconn setting
set rate-limit : change a rate limiting value
set severity-output [none|number|string] : set presence of severity level in feedback information
set timeout    : change a timeout setting
show env [var] : dump environment variables known to the process
show cli sockets : dump list of cli sockets
show cli level   : display the level of the current CLI session
show fd [num] : dump list of file descriptors in use
show activity : show per-thread activity stats (for support/developers)
operator       : lower the level of the current CLI session to operator
user           : lower the level of the current CLI session to user
clear table    : remove an entry from a table
set table [id] : update or create a table entry's data
show table [id]: report table usage stats or dump this table's contents
disable frontend : temporarily disable specific frontend
enable frontend : re-enable specific frontend
set maxconn frontend : change a frontend's maxconn setting
show servers state [id]: dump volatile server information (for backend <id>)
show backend   : list backends in the current running config
shutdown frontend : stop a specific frontend
set dynamic-cookie-key backend : change a backend secret key for dynamic cookies
enable dynamic-cookie backend : enable dynamic cookies on a specific backend
disable dynamic-cookie backend : disable dynamic cookies on a specific backend
show errors    : report last request and response errors for each proxy
show resolvers [id]: dumps counters from all resolvers section and
associated name servers
show cache     : show cache status
add acl        : add acl entry
clear acl <id> : clear the content of this acl
del acl        : delete acl entry
get acl        : report the patterns matching a sample for an ACL
show acl [id]  : report available acls or dump an acl's contents
add map        : add map entry
clear map <id> : clear the content of this map
del map        : delete map entry
get map        : report the keys and values matching a sample for a map
set map        : modify map entry
show map [id]  : report available maps or dump a map's contents
trace <module> [cmd [args...]] : manage live tracing
show trace [<module>] : show live tracing state
show threads   : show some threads debugging information
show pools     : report information about the memory pools usage
show events [<sink>] : show event sink state
show profiling : show CPU profiling options
set  profiling : enable/disable CPU profiling
[root@centos7 ~]#echo "show info" | socat stdio /var/lib/haproxy/haproxy.sock
Name: HAProxy
Version: 2.1.3
Release_date: 2020/02/12
Nbthread: 4
Nbproc: 1
Process_num: 1
Pid: 2279
Uptime: 0d 0h46m07s
Uptime_sec: 2767
Memmax_MB: 0
PoolAlloc_MB: 0
PoolUsed_MB: 0
PoolFailed: 0
Ulimit-n: 200041
Maxsock: 200041
Maxconn: 100000
Hard_maxconn: 100000
CurrConns: 0
CumConns: 1
CumReq: 1
MaxSslConns: 0
CurrSslConns: 0
CumSslConns: 0
Maxpipes: 0
PipesUsed: 0
PipesFree: 0
ConnRate: 0
ConnRateLimit: 0
MaxConnRate: 0
SessRate: 0
SessRateLimit: 0
MaxSessRate: 0
SslRate: 0
SslRateLimit: 0
MaxSslRate: 0
SslFrontendKeyRate: 0
SslFrontendMaxKeyRate: 0
SslFrontendSessionReuse_pct: 0
SslBackendKeyRate: 0
SslBackendMaxKeyRate: 0
SslCacheLookups: 0
SslCacheMisses: 0
CompressBpsIn: 0
CompressBpsOut: 0
CompressBpsRateLim: 0
ZlibMemUsage: 0
MaxZlibMemUsage: 0
Tasks: 19
Run_queue: 1
Idle_pct: 100
node: centos7.wangxiaochun.com
Stopping: 0
Jobs: 7
Unstoppable Jobs: 0
Listeners: 6
ActivePeers: 0
ConnectedPeers: 0
DroppedLogs: 0
BusyPolling: 0
FailedResolutions: 0
TotalBytesOut: 0
BytesOutRate: 0
DebugCommandsIssued: 0
[root@centos7 ~]#cat /etc/haproxy/haproxy.cfg
......
listen magedu-test-80
bind :81,:82
mode http
server web1 10.0.0.17:80  check inter 3000 fall 3 rise 5 
server web2 10.0.0.27:80 check weight 3  
......
[root@centos7 ~]#echo "show servers state" | socat stdio /var/lib/haproxy/haproxy.sock
1
# be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id srv_fqdn srv_port srvrecord
2 magedu-test-80 1 web1 10.0.0.17 2 0 2 1 812 6 3 7 6 0 0 0 - 80 -
2 magedu-test-80 2 web2 10.0.0.27 2 0 2 3 812 6 3 4 6 0 0 0 - 80 -
4 web_port 1 web1 127.0.0.1 0 0 1 1 810 8 2 0 6 0 0 0 - 8080 -
[root@centos7 ~]#echo "get weight magedu-test-80/web2" | socat stdio /var/lib/haproxy/haproxy.sock
3 (initial 3)
#修改weight,注意只针对单进程有效
[root@centos7 ~]#echo "set weight magedu-test-80/web2 2" | socat stdio /var/lib/haproxy/haproxy.sock
[root@centos7 ~]#echo "get weight magedu-test-80/web2" | socat stdio /var/lib/haproxy/haproxy.sock
2 (initial 3)
#将后端服务器禁用,注意只针对单进程有效
[root@centos7 ~]#echo "disable server magedu-test-80/web2" | socat stdio /var/lib/haproxy/haproxy.sock
#将后端服务器软下线,即weight设为0
[root@centos7 ~]#echo "set weight magedu-test-80/web1 0" | socat stdio /var/lib/haproxy/haproxy.sock
#将后端服务器禁用,针对多进程
[root@centos7 ~]#vim /etc/haproxy/haproxy.cfg
......
stats socket /var/lib/haproxy/haproxy1.sock mode 600 level admin process 1
stats socket /var/lib/haproxy/haproxy2.sock mode 600 level admin process 2               nbproc 2
.....
[root@centos7 ~]#echo "disable server  magedu-test-80/web2" | socat stdio /var/lib/haproxy/haproxy1.sock
[root@centos7 ~]#echo "disable server  magedu-test-80/web2" | socat stdio /var/lib/haproxy/haproxy2.sock
[root@haproxy ~]#for i in {1..2};do echo "set weight magedu-test-80/webi 10" | socat stdio /var/lib/haproxy/haproxyi.sock;done
#如果静态算法,如:static-rr,可以更改weight为0或1,但不支持动态更改weight为其它值,否则会提示下面信息
[root@centos7 ~]#echo "set weight magedu-test-80/web1 0" | socat stdio /var/lib/haproxy/haproxy.sock
[root@centos7 ~]#echo "set weight magedu-test-80/web1 1" | socat stdio /var/lib/haproxy/haproxy.sock
[root@centos7 ~]#echo "set weight magedu-test-80/web1 2" | socat stdio /var/lib/haproxy/haproxy.sock
Backend is using a static LB algorithm and only accepts weights '0%' and '100%'.

static-rr

static-rr:基于权重的轮询调度,不支持权重的运行时利用socat进行动态调整及后端服务器慢启动,其后端主机数量没有限制,相当于LVS中的 wrr

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance static-rr
server web1  10.0.0.17:80 weight 1 check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 2 check inter 3000 fall 2 rise 5

first

first:根据服务器在列表中的位置,自上而下进行调度,但是其只会当第一台服务器的连接数达到上限,新请求才会分配给下一台服务,因此会忽略服务器的权重设置,此方式使用较少

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance first
server web1  10.0.0.17:80 maxconn 2 weight 1 check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1 check inter 3000 fall 2 rise 5

测试访问效果

#同时运行下面命令,观察结果
# while  true;do  curl http://10.0.0.7/index.html ; sleep 0.1;done

动态算法

动态算法:基于后端服务器状态进行调度适当调整,优先调度至当前负载较低的服务器,且权重可以在haproxy运行时动态调整无需重启。

roundrobin

roundrobin:基于权重的轮询动态调度算法,支持权重的运行时调整,不同于lvs中的rr轮训模式,HAProxy中的roundrobin支持慢启动(新加的服务器会逐渐增加转发数),其每个后端backend中最多支持4095个real server,支持对real server权重动态调整,roundrobin为默认调度算法

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance roundrobin
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 2  check inter 3000 fall 2 rise 5

支持动态调整权重:

# echo "get weight web_host/web1" | socat stdio /var/lib/haproxy/haproxy.sock 
1 (initial 1)
# echo "set weight web_host/web1 3" | socat stdio /var/lib/haproxy/haproxy.sock 
# echo "get weight web_host/web1" | socat stdio /var/lib/haproxy/haproxy.sock 
3 (initial 1)

leastconn

leastconn加权的最少连接的动态,支持权重的运行时调整和慢启动,即当前后端服务器连接最少的优先调度(新客户端连接),比较适合长连接的场景使用,比如:MySQL等场景。

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance leastconn
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

random

在1.9版本开始增加一个叫做random的负载平衡算法,其基于随机数作为一致性hash的key,随机负载平衡对于大型服务器场或经常添加或删除服务器非常有用,支持weight的动态调整,weight较大的主机有更大概率获取新请求

random配置实例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance  random
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

其他算法

其它算法即可作为静态算法,又可以通过选项成为动态算法

source

源地址hash,基于用户源地址hash并将请求转发到后端服务器,后续同一个源地址请求将被转发至同一个后端web服务器。此方式当后端服务器数据量发生变化时,会导致很多用户的请求转发至新的后端服务器,默认为静态方式,但是可以通过hash-type支持的选项更改

这个算法一般是在不插入Cookie的TCP模式下使用,也可给拒绝会话cookie的客户提供最好的会话粘性,适用于session会话保持但不支持cookie和缓存的场景

源地址有两种转发客户端请求到后端服务器的服务器选取计算方式,分别是取模法和一致性hash

map-base取模法

map-based:取模法,对source地址进行hash计算,再基于服务器总权重的取模,最终结果决定将此请求转发至对应的后端服务器。此方法是静态的,即不支持在线调整权重,不支持慢启动,可实现对后端服务器均衡调度。缺点是当服务器的总权重发生变化时,即有服务器上线或下线,都会因总权重发生变化而导致调度结果整体改变,hash-type 指定的默认值为此算法

所谓取模运算,就是计算两个数相除之后的余数,10%7=3, 7%4=3
map-based算法:基于权重取模,hash(source_ip)%所有后端服务器相加的总权重

取模法配置示例:

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode tcp
log global
balance source
hash-type map-based 
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 3
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 3
[root@haproxy ~]#echo "set weight web_host/10.0.0.27 10" | socat stdio /var/lib/haproxy/haproxy.sock 
Backend is using a static LB algorithm and only accepts weights '0%' and '100%'.
[root@haproxy ~]#echo "set weight web_host/10.0.0.27 0" | socat stdio /var/lib/haproxy/haproxy.sock 
[root@haproxy conf.d]#echo "get weight web_host/10.0.0.27" | socat stdio /var/lib/haproxy/haproxy.sock 
0 (initial 1)

一致性hash

一致性哈希,当服务器的总权重发生变化时,对调度结果影响是局部的,不会引起大的变动,hash(o)mod n ,该hash算法是动态的,支持使用 socat等工具进行在线权重调整,支持慢启动

算法:

1、key1=hash(source_ip)%(2^32)  [0---4294967295]
2、keyA=hash(后端服务器虚拟ip)%(2^32)
3、将key1和keyA都放在hash环上,将用户请求调度到离key1最近的keyA对应的后端服务器

hash环偏斜问题

增加虚拟服务器IP数量,比如:一个后端服务器根据权重为1生成1000个虚拟IP,再hash。而后端服务器权重为2则生成2000的虚拟IP,再bash,最终在hash环上生成3000个节点,从而解决hash环偏斜问题
hash对象

Hash对象到后端服务器的映射关系:

Haproxy-调度算法详解插图

一致性hash示意图

后端服务器在线与离线的调度方式

Haproxy-调度算法详解插图(1)
Haproxy-调度算法详解插图(2)

一致性hash配置示例
listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode tcp
log global
balance source
hash-type consistent
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

uri

基于对用户请求的URI的左半部分或整个uri做hash,再将hash结果对总权重进行取模后,根据最终结果将请求转发到后端指定服务器,适用于后端是缓存服务器场景,默认是静态,也可以通过hash-type指定map-based和consistent,来定义使用取模法还是一致性hash。

注意:此算法是应用层,所有只支持 mode http ,不支持 mode tcp

<scheme>://<user>:<password>@<host>:<port>/<path>;<params>?<query>#<frag>
左半部分:/<path>;<params>
整个uri:/<path>;<params>?<query>#<frag>

Haproxy-调度算法详解插图(3)

uri 取模法配置示例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance uri
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

uri 一致性hash配置示例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance uri
hash-type consistent
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

访问测试

访问不同的uri,确认可以将用户同样的请求转发至相同的服务器

# curl  http://10.0.0.7/test1.html
# curl  http://10.0.0.7/test2..html

url_param

url_param对用户请求的url中的 params 部分中的一个参数key对应的value值作hash计算,并由服务器总权重相除以后派发至某挑出的服务器;通常用于追踪用户,以确保来自同一个用户的请求始终发往同一个real server,如果无没key,将按roundrobin算法

假设:
url = http://www.magedu.com/foo/bar/index.php?key=value
则:
host = "www.magedu.com"
url_param = "key=value"

url_param取模法配置示例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance url_param  userid     #url_param hash
server web1 10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2 10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

url_param一致性hash配置示例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance url_param  userid             #对url_param的值取hash
hash-type consistent
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

测试访问

# curl  http://10.0.0.7/index.html?userid=<NAME_ID> 
# curl  "http://10.0.0.7/index.html?userid=<NAME_ID>&typeid=<TYPE_ID>" 

hdr

针对用户每个http头部(header)请求中的指定信息做hash,此处由 name 指定的http首部将会被取出并做hash计算,然后由服务器总权重取模以后派发至某挑出的服务器,如无有效的值,则会使用默认的轮询调度。

hdr取模法配置示例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance hdr(User-Agent)
#balance hdr(host)
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

一致性hash配置示例

listen  web_host
bind 10.0.0.7:80,:8801-8810,10.0.0.7:9001-9010
mode http
log global
balance hdr(User-Agent)
hash-type consistent
server web1  10.0.0.17:80 weight 1  check inter 3000 fall 2 rise 5
server web2  10.0.0.27:80 weight 1  check inter 3000 fall 2 rise 5

测试访问

[root@centos6 ~]#curl -v http://10.0.0.7/index.html
[root@centos6 ~]#curl -vA 'firefox' http://10.0.0.7/index.html
[root@centos6 ~]#curl -vA 'chrome' http://10.0.0.7/index.html

rdp-cookie

rdp-cookie对远windows远程桌面的负载,使用cookie保持会话,默认是静态,也可以通过hash-type指定map-based和consistent,来定义使用取模法还是一致性hash。

rdp-cookie取模法配置示例

listen RDP
bind 10.0.0.7:3389
balance rdp-cookie
mode tcp
server rdp0 10.0.0.17:3389 check fall 3 rise 5 inter 2000 weight 1

rdp-cookie一致性hash配置示例

[root@haproxy ~]#cat /etc/haproxy/conf.d/windows_rdp.cfg 
listen magedu_RDP_3389
bind  172.16.0.100:3389
balance rdp-cookie
hash-type consistent
mode tcp
server rdp0 10.0.0.200:3389 check fall 3 rise 5 inter 2000 weight 1
[root@haproxy ~]#hostname -I
10.0.0.100 172.16.0.100 

Haproxy-调度算法详解插图(4)
Haproxy-调度算法详解插图(5)

基于iptables实现RDP协议转发

必须开启ip转发功能:net.ipv4.ip_forward = 1

[root@centos8 ~]#sysctl -w net.ipv4.ip_forward=1
#客户端和Windows在不同网段需要下面命令
[root@centos8 ~]#iptables  -t nat -A PREROUTING -d 172.16.0.100 -p tcp --dport 3389 -j DNAT --to-destination 10.0.0.200:3389
#客户端和Windows在同一网段需要再执行下面命令
[root@centos8 ~]#iptables  -t nat -A PREROUTING -d 10.0.0.8 -p tcp --dport 3389 -j DNAT --to-destination 10.0.0.1:3389
[root@centos8 ~]#iptables -t nat -A POSTROUTING -s 10.0.0.0/24 -j SNAT --to-source  10.0.0.8

在windows 可以看到以下信息

Haproxy-调度算法详解插图(6)

算法总结

static-rr--------->tcp/http  静态
first------------->tcp/http  静态
roundrobin-------->tcp/http 动态
leastconn--------->tcp/http 动态
random------------>tcp/http 动态
以下静态和动态取决于hash_type是否consistent
source------------>tcp/http
Uri--------------->http
url_param--------->http     
hdr--------------->http
rdp-cookie-------->tcp

各算法使用场景

first       #使用较少
static-rr   #做了session共享的web集群
roundrobin
random
leastconn   #数据库
source      #基于客户端公网IP的会话保持
Uri--------------->http  #缓存服务器,CDN服务商,蓝汛、百度、阿里云、腾讯
url_param--------->http 
hdr         #基于客户端请求报文头部做下一步处理
rdp-cookie  #很少使用

本文链接:http://www.yunweipai.com/35268.html

原创文章,作者:ItWorker,如若转载,请注明出处:https://blog.ytso.com/52698.html

(0)
上一篇 2021年8月6日
下一篇 2021年8月6日

相关推荐

发表回复

登录后才能评论