Linux Cluster with HAProxy and Keepalived - Revision history

Pio2pio: /* Install */

2019-11-19T22:59:39Z

Install

← Older revision		Revision as of 00:59, 20 November 2019
Line 77:		Line 77:
	icmp-blocks:		icmp-blocks:
	rich rules:		rich rules:

			# add permanent ingress rule to allow port 8080
			sudo firewall-cmd --permanent --add-port=8080/tcp
			sudo firewall-cmd --reload
			...
			ports: 8080/tcp
			...
	</source>		</source>

Pio2pio: /* Configuration File */

2019-11-19T22:39:28Z

Configuration File

← Older revision		Revision as of 00:39, 20 November 2019
Line 96:		Line 96:
	daemon		daemon
	maxconn 256		maxconn 256
			#---------------------------------------------------------------------
			# common defaults that all the 'listen' and 'backend' sections will
			# use if not designated in their block
			#---------------------------------------------------------------------
	defaults		defaults
	mode http		mode http
Line 104:		Line 107:
	option forwardfor		option forwardfor
	option http-server-close		option http-server-close
			#---------------------------------------------------------------------
			# main frontend which proxys to the backends
			#---------------------------------------------------------------------
	frontend http-in		frontend http-in
	bind *:80		bind *:80
	rspadd X-Forwarded-Host:\ http:\\\\example.com		rspadd X-Forwarded-Host:\ http:\\\\example.com
	default_backend appZ-backend		default_backend appZ-backend

	backend appZ-backend		backend appZ-backend
	balance roundrobin		balance roundrobin
Line 122:		Line 127:
	As mentioned previously, HAProxy has the ability to load balance using layer 4 or 7 in the OSI model. With that in mind and by setting the mode to http, HAProxy can now inspect HTTP headers for all requests and modify and redirect per each request. The other option is to use tcp as the mode which allows HAProxy to route per IP and port; ignoring all HTTP headers.		As mentioned previously, HAProxy has the ability to load balance using layer 4 or 7 in the OSI model. With that in mind and by setting the mode to http, HAProxy can now inspect HTTP headers for all requests and modify and redirect per each request. The other option is to use tcp as the mode which allows HAProxy to route per IP and port; ignoring all HTTP headers.

	Reference: https://cbonte.github.io/haproxy-dconv/1.6/configuration.html#mode		Reference: https://cbonte.github.io/haproxy-dconv/1.6/configuration.html#mode

	=== timeout 's ===		=== timeout 's ===
	HAProxy has the ability to specify timeouts for various aspects of standard HTTP/TCP protocols. In the example, the configuration file has been configured to limit connection (time to wait for a successful connection attempt), client (maximum client side inactivity time) and server (maximum server side inactivity time) time frames.		HAProxy has the ability to specify timeouts for various aspects of standard HTTP/TCP protocols. In the example, the configuration file has been configured to limit connection (time to wait for a successful connection attempt), client (maximum client side inactivity time) and server (maximum server side inactivity time) time frames.

Pio2pio: /* Structure */

2019-11-19T22:33:56Z

Structure

@@ Line 55: / Line 55: @@
 === Least Connection ===
 Each resource in a given backend is evaluated to determine which one has the least number of active connections. The resource with the lowest number will receive the next request. The developers of HAProxy state that this algorithm is a great option for connections that are expected to last a long time such as LDAP and SQL but not for HTTP.
 == Structure ==
 === Files & Folders ===
@@ Line 114: / Line 138: @@
 === rspadd X-Forwarded-Host:\===
 As the forwardfor option enables the use of the X-Forwarded-For header, it must now be configured and that is what rspadd is used for. One of the more common uses of this is when hosting various applications behind HAProxy and you want to ensure that your end users only see responses coming from HAProxy and not the machines behind it which can occasionally occur with Asynchronous processes.
 == Usage ==
 To bring the above information into further context, below are a few basic example configurations to help illustrate how one would setup HAProxy for different scenarios.

Pio2pio: /* Troubleshooting */

2019-11-18T22:17:10Z

Troubleshooting

← Older revision		Revision as of 00:17, 19 November 2019
Line 369:		Line 369:

	== Troubleshooting ==		== Troubleshooting ==
	[root@dc1server01 ~]# fgrep -i keep /var/log/messages*		[root@dc1server01 ~]# fgrep -i keep /var/log/messages*

	/var/log/messages-20160822:Aug 22 09:30:58 dc1server01 Keepalived[1141]: Starting Keepalived v1.2.13 (11/18,2014)		/var/log/messages-20160822:Aug 22 09:30:58 dc1server01 Keepalived[1141]: Starting Keepalived v1.2.13 (11/18,2014)
	/var/log/messages-20160822:Aug 22 09:30:58 dc1server01 Keepalived[1563]: Starting Healthcheck child process, pid=1564		/var/log/messages-20160822:Aug 22 09:30:58 dc1server01 Keepalived[1563]: Starting Healthcheck child process, pid=1564
Line 404:		Line 403:
	/var/log/messages-20160822:Aug 22 09:38:35 dc1server01 Keepalived_vrrp[1565]: VRRP_Instance(dc1server) Sending gratuitous ARPs on eth0 for 10.0.10.3		/var/log/messages-20160822:Aug 22 09:38:35 dc1server01 Keepalived_vrrp[1565]: VRRP_Instance(dc1server) Sending gratuitous ARPs on eth0 for 10.0.10.3
	/var/log/messages-20160822:Aug 22 09:44:53 dc1server01 Keepalived_vrrp[1565]: VRRP_Instance(dc1server) Received lower prio advert, forcing new election		/var/log/messages-20160822:Aug 22 09:44:53 dc1server01 Keepalived_vrrp[1565]: VRRP_Instance(dc1server) Received lower prio advert, forcing new election


	[root@dc2server01 ~]# fgrep -i keep /var/log/messages*		[root@dc2server01 ~]# fgrep -i keep /var/log/messages*

Pio2pio: /* Troubleshooting */

2019-11-18T22:16:40Z

Troubleshooting

← Older revision		Revision as of 00:16, 19 November 2019
Line 412:		Line 412:
	/var/log/messages:Aug 22 10:53:05 dc2server01 Keepalived_healthcheckers[1233]: Netlink reflector reports IP 10.0.20.3 removed		/var/log/messages:Aug 22 10:53:05 dc2server01 Keepalived_healthcheckers[1233]: Netlink reflector reports IP 10.0.20.3 removed
	/var/log/messages:Aug 22 11:04:49 dc2server01 Keepalived_vrrp[1234]: VRRP_Script(chk_haproxy) succeeded		/var/log/messages:Aug 22 11:04:49 dc2server01 Keepalived_vrrp[1234]: VRRP_Script(chk_haproxy) succeeded


			;Multiline config
			You can't do multiline syntax in the haproxy.cfg [https://cbonte.github.io/haproxy-dconv/1.8/configuration.html#2.1 defined format documentation]

Pio2pio: /* Keepalived */

2019-11-18T22:14:25Z

Keepalived

Show changes

Pio2pio: /* HAproxy */

2019-11-18T22:06:51Z

HAproxy

Show changes

Pio2pio: /* Usage */

2019-11-18T21:48:43Z

Usage

← Older revision		Revision as of 23:48, 18 November 2019
Line 143:		Line 143:
	server appZ_02 192.168.2.5:8080 check backup		server appZ_02 192.168.2.5:8080 check backup
	</source>		</source>
			== [https://linuxacademy.com/guide/14569-configuration-of-haproxy/ Configuration of HAProxy] ==

	= Application Stack With HAproxy, Keepalived and postfix example =		= Application Stack With HAproxy, Keepalived and postfix example =

Pio2pio: /* HAproxy guide */

2018-06-19T18:43:08Z

HAproxy guide

← Older revision		Revision as of 20:43, 19 June 2018
Line 13:		Line 13:

	Example frontend:		Example frontend:
	<~~script~~>		<source>
	frontend http-in		frontend http-in
	bind *:80		bind *:80
	default_backend appX-backend		default_backend appX-backend
	</~~script~~>		</source>
	This frontend example shows that we are only opening port 80 for incoming requests and redirecting all traffic to the default backend called appX-backend.		This frontend example shows that we are only opening port 80 for incoming requests and redirecting all traffic to the default backend called appX-backend.
	=== Backend ===		=== Backend ===
Line 27:		Line 27:

	Example backend:		Example backend:
	<~~script~~>		<source>
	backend appX-backend		backend appX-backend
	balance roundrobin		balance roundrobin
	server appX_01 192.168.2.2:8080 check		server appX_01 192.168.2.2:8080 check
	server appX_02 192.168.2.3:8080 check		server appX_02 192.168.2.3:8080 check
	</~~script~~>		</source>
	In this example, the backend is called appX-backend and it contains two servers that are being accessed using roundrobin which is explained later in this guide.		In this example, the backend is called appX-backend and it contains two servers that are being accessed using roundrobin which is explained later in this guide.

Line 41:		Line 41:

	Example ACL:		Example ACL:
	<~~script~~>		<source>
	frontend http-in		frontend http-in
	bind *:80		bind *:80
Line 47:		Line 47:
	use_backend appX-backend if url_appX		use_backend appX-backend if url_appX
	default_backend appZ-backend		default_backend appZ-backend
	</~~script~~>		</source>
	This example is rather simple in that it evaluates the incoming request for the resource path which is the context immediately after the first / such as http://example.com/appX. In this scenario, if the incoming request is http://example.com/appX, the request is sent to the backend called appX-backend whereas all other requests will default to the appZ-backend.		This example is rather simple in that it evaluates the incoming request for the resource path which is the context immediately after the first / such as http://example.com/appX. In this scenario, if the incoming request is http://example.com/appX, the request is sent to the backend called appX-backend whereas all other requests will default to the appZ-backend.
	=== Algorithms ===		=== Algorithms ===

Pio2pio: /* HAproxy guide */

2018-06-19T18:41:34Z

HAproxy guide

Show changes