HAProxy Enterprise Documentation 2.5r1

Discovery with SRV Records

DNS SRV records specify the host and port that a service listens on. You can configure HAProxy Enterprise to query for these records and populate the IP addresses and ports of server lines in a backend section.

[DNS SRV records]

About DNS SRV records

DNS SRV records are resources used to identify computers that host specific services. They are contained in the ANSWER section of DNS responses and have the following structure:

_service._proto.name. TTL class SRV priority weight port target



Standard network service name (taken from /etc/services) or a port number


Standard protocol name ("tcp" or "udp")


Name of the service, i.e. the name used in the query


Validity period for the response (HAProxy Enterprise ignores this field because it maintains its own expiry data defined in the configuration)


DNS class ("IN")


DNS record type ("SRV")


Priority of the target host. Lower value = higher preference (HAProxy Enterprise ignores this field but may use it later to indicate active / backup state)


Relative weight in case of records with the same priority. Higher number = higher preference


Port where the service is configured


Hostname of the machine providing the service, ending in a dot

Configure service discovery

  1. Update your DNS nameserver to resolve a service name, such as myservice.example.local, to one or more hostnames and ports via SRV records. Those hostnames should resolve to IP addresses using A records.

    • Add DNS A records that resolve multiple hostnames, such as host1, host2 and host3, to different IP addresses.

    • Add the same number of SRV records that resolve a service name, such as _myservice._tcp.example.local, to the hostnames you defined and the port on the host where the service listens.

  2. Query the nameserver directly with the dig tool to ensure that it returns the correct records, as follows:

    $ dig @ -p 53 SRV _myservice._tcp.example.local
       ;_myservice._tcp.example.local. IN
       _myservice._tcp.example.local. 0 IN     SRV     0 0 8080 host1.
       _myservice._tcp.example.local. 0 IN     SRV     0 0 8081 host2.
       _myservice._tcp.example.local. 0 IN     SRV     0 0 8082 h
       host1.                  0       IN      A
       host2.                  0       IN      A
       host3.                  0       IN      A
  3. Add a resolvers section in your HAProxy Enterprise configuration file to set the DNS nameservers to watch for changes.

    • Add one or more nameserver lines to specify the IP addresses and ports of your DNS nameservers.

    • Set the accepted_payload_size to 8192 to allow larger DNS payloads, which is required to receive more server IPs within a single DNS result.

    resolvers mydns
       nameserver dns1
       accepted_payload_size 8192
  4. Use a server-template in a backend to set the template for the server lines when HAProxy Enterprise queries your DNS servers, as follows:

    backend webservers
       balance roundrobin
       server-template web 5 _myservice._tcp.example.local resolvers mydns check init-addr none

    In this example, the server-template directive:

    • Adds the specified number of servers (5) to the backend.

    • Appends "web" as a prefix to their names.

    • Queries the service name _myservice._tcp.example.local.

    • Have the SRV records fill in the ports.

    • Specifies the resolvers section mydns

    • The init-addr none argument means that HAProxy Enterprise can initialize without having to resolve the IP addresses at startup. It can resolve them during runtime.

    This is equivalent to adding a backend to HAProxy Enterprise that looks like this:

    backend webservers
        balance roundrobin
        server web1 check
        server web2 check
        server web3 check
        server web4 check disabled
        server web5 check disabled

    When you add more records to your nameserver, they will automatically go into the backend to fill in the web4 and web5 slots.

Inspect servers with the Runtime API

Run the command show servers state [backend name] to see the servers loaded into memory:

$ echo "show servers state webservers" | sudo socat stdio unix-connect:/var/run/hapee-2.5/hapee-lb.sock

  # be_id  be_name  srv_id  srv_name  srv_addr       srv_fqdn  srv_port  srvrecord
  3        servers  1       web1  host3     8082      _myservice._tcp.example.local
  3        servers  2       web2  host2     8081      _myservice._tcp.example.local
  3        servers  3       web3  host1     8080      _myservice._tcp.example.local
  3        servers  4       web4      -              -         0         _myservice._tcp.example.local
  3        servers  5       web5      -              -         0         _myservice._tcp.example.local

Next up