HAProxy Configuration
Internal
Overview
If installed with yum, the default configuration file is deployed in /etc/haproxy/haproxy.cfg and the systemd configuration file in /etc/sysconfig/haproxy.
HAProxy's configuration process involves 3 sources of configuration parameters:
- Arguments from the command line, which always take precedence over file configuration.
- The global section, which sets process-wide parameters.
- The proxies sections, which are defaults, listen, frontend and backend.
TODO
HAProxy and OpenShift: HAProxy on lb.local. It seems that if none of the masters are not available when HAProxy starts, it fails and then it does not retry. How to I configure it to retry for each request? Maybe Ansible knows how to do that?
Example
The following example proxies HTTPS connections by passing them directly to the backend.
#---------------------------------------------------------------------
# Global settings
#---------------------------------------------------------------------
global
maxconn 20000
log 127.0.0.1:514 local2
chroot /var/lib/haproxy
pidfile /var/run/haproxy.pid
user haproxy
group haproxy
daemon
# turn on stats unix socket
stats socket /var/lib/haproxy/stats
#---------------------------------------------------------------------
# common defaults that all the 'listen' and 'backend' sections will
# use if not designated in their block
#---------------------------------------------------------------------
defaults
mode http
log global
option httplog
option dontlognull
# option http-server-close
# option forwardfor except 127.0.0.0/8
option redispatch
retries 3
timeout http-request 10s
timeout queue 1m
timeout connect 10s
timeout client 300s
timeout server 300s
timeout http-keep-alive 10s
timeout check 10s
maxconn 20000
listen stats :9000
mode http
stats enable
stats uri /
frontend atomic-openshift-api
bind *:443
mode tcp
option tcplog
default_backend atomic-openshift-api
backend atomic-openshift-api
mode tcp
balance source
server master1 192.168.122.13:443 check
server master2 192.168.122.14:443 check
server master3 192.168.122.15:443 check
Dependency on Other Services
Under some circumstances, HAProxy need other services to start before it starts, so it can rely on them. For example, if a local DNS server resolves the names referred from HAProxy configuration file, the named service must start before HAProxy. This is configured in the HAProxy's unit file /usr/lib/systemd/system/haproxy.service:
Requires=named.service After=syslog.target network.target named.service
More details: Declaring a Dependency on a Service.
Logging Configuration
Logging Destination Configuration
HAProxy logging concepts:
Logging configuration consists of the following steps:
Add the following to the "global" section of the configuration file:
log 127.0.0.1:514 local2
and then add the following to each "defaults" section or to each frontend and backend section:
log global
Then make sure the local syslogd does listen to the UDP traffic. For details on how to do this for rsyslogd, see:
Configure HAProxy to Log into a File
Assuming that logging was configured as described in the previous section, configure local2 events to go to the /var/log/haproxy.log file. Add the following line in /etc/rsyslog.conf:
local2.* /var/log/haproxy.log
Log Format
HAProxy comes with two pre-defined log formats:
1. HTTP log format:
%ci:%cp\ [%tr]\ %ft\ %b/%s\ %TR/%Tw/%Tc/%Tr/%Ta\ %ST\ %B\ %CC\ %CS\ %tsc\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq\ %hr\ %hs\ %{+Q}r
2. TCP log format:
%ci:%cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq
These can be requested in the corresponding section ("listen", "fronted", etc.) with the following declaration:
... frontend ... option tcplog|httplog
The log format can be customized, by declaring it in "defaults" section or in the appropriate "listen" or "frontend"
log-format %Ci:%Cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq
Important: When declaring a custom log format, spaces must be escaped ('\ '), because log-format expects just one argument.
Also important: When using a custom log format, options "tcplog" or "httplog" in the "defaults" or "frontend" sections must be commented out, otherwise they take precedence and the custom format does not surface.
SSL Configuration
Subjects
Configuration Reference
Options
httplog
tcplog
Logging is set to tcp instead of the default http.
ssl-hello-chk
A health check that verifies the the connection and its ability to handle SSL (SSLv3 specifically) connections.
global
The "global" section sets process-wide parameters.
defaults
This is one of four "proxy" sections (defaults, listen, frontend, backend). It sets default parameters for all other sections following its declaration. These default parameters are overwritten by the subsequent "defaults" sections, if present.
mode
Possible values:
http
tcp
Used to pass secure connections off to a backend server without encrypting it.
balance
Specifies the algorithm used to select a server when doing load balancing. This only applies when no persistence information is available, or when a connection is redispatched to another server. Possible values:
roundrobin
Each server is used in turns, according to their weights. This is the smoothest and fairest algorithm when the server's processing time remains equally distributed. This algorithm is dynamic, which means that server weights may be adjusted on the fly for slow starts for instance. It is limited by design to 4095 active servers per backend.
source
The source IP address is hashed and divided by the total weight of the running servers to designate which server will receive the request. This ensures that the same client IP address will always reach the same server as long as no server goes down or up. If the hash result changes due to the number of running servers changing, many clients will be directed to a different server. This algorithm is generally used in TCP mode where no cookie may be inserted. It may also be used on the Internet to provide a best-effort stickiness to clients which refuse session cookies. This algorithm is static by default, which means that changing a server's weight on the fly will have no effect, but this can be changed using "hash-type".
static-rr
leastconn
first
uri
url_param
hdr(name)
rdp-cookie, rdp-cookie(name)
listen
This is one of four "proxy" sections (defaults, listen, frontend, backend). It defines a complete proxy with its fronted and backend parts combined in one section. It is generally useful for TCP-only traffic.
frontent
This is one of four "proxy" sections (defaults, listen, frontend, backend). It describes a set of listening sockets accepting client connections.
backend
This is one of four "proxy" sections (defaults, listen, frontend, backend). It describes a set of servers to which the proxy will connect to forward incoming connections.