Agent Proxy Configuration

Overview

If your network configuration restricted outbound traffic, proxy all Agent traffic through one or several hosts that have more permissive outbound policies.

A few options are available to send traffic to Datadog over SSL/TLS for hosts that are not directly connected to the Internet.

  1. Using a web proxy, such as Squid or Microsoft Web Proxy, that is already deployed to your network
  2. Using HAProxy (if you want to proxy more than 16-20 Agents through the same proxy)
  3. Using the Agent as a proxy (for up to 16 Agents per proxy, only on Agent v5 )

Web proxy

For specific information regarding Squid, see the Squid section of this page.

Traditional web proxies are supported natively by the Agent. If you need to connect to the Internet through a proxy, edit your Agent configuration file.

Agent v6 & v7

Set different proxy servers for https and http requests in your Agent datadog.yaml configuration file. The Agent uses https to send data to Datadog, but integrations might use http to gather metrics. No matter the proxied requests, you can activate SSL on your proxy server. Below are some configuration examples for your datadog.yaml file.

If log collection is enabled, make sure that a specific transport is enforced. The recommended setup is to use HTTPS. In that case, the <HOST>:<PORT> used to proxy metrics is used to proxy logs. If you are using TCP transport, see TCP Proxy for Logs.

Setting an HTTP proxy for all https requests:

proxy:
    https: "http://<SSL_PROXY_SERVER_FOR_HTTPS>:<PORT>"

Note: When setting up an HTTP proxy for https requests, the actual communication between the Agent and Datadog is encrypted end-to-end with TLS and cannot be decrypted by the proxy. The only unencrypted communication is the HTTP CONNECT request that’s made between the Agent and the proxy to establish the initial TCP connection between the Agent and Datadog. As such, when using a proxy for https requests, there is no need to use an HTTPS proxy in order to have encrypted communication between the Agent and Datadog.

Setting an HTTPS proxy for both https and http requests:

proxy:
    https: "https://<SSL_PROXY_SERVER_FOR_HTTPS>:<PORT>"
    http: "https://<SSL_PROXY_SERVER_FOR_HTTP>:<PORT>"

Setting a <USERNAME> and <PASSWORD> to contact the proxy server for both https and http requests:

proxy:
    https: "http://<USERNAME>:<PASSWORD>@<PROXY_SERVER_FOR_HTTPS>:<PORT>"
    http: "http://<USERNAME>:<PASSWORD>@<PROXY_SERVER_FOR_HTTP>:<PORT>"

Using the no_proxy list to specify hosts that must bypass the proxy:

proxy:
    https: "http://<USERNAME>:<PASSWORD>@<PROXY_SERVER_FOR_HTTPS>:<PORT>"
    http: "http://<USERNAME>:<PASSWORD>@<PROXY_SERVER_FOR_HTTP>:<PORT>"
    no_proxy:
      - host1
      - host2

Note: All integrations that make HTTP(S) requests default back to proxy settings defined in datadog.yaml configuration file if none are specified at the integration level. If this is undesired, set skip_proxy to true or use_agent_proxy to false in every instance config or in the init_config fallback for your integration.

NO_PROXY accepted values

By default, no_proxy/NO_PROXY must match endpoints exactly for Agent HTTP(S) requests (except requests performed by Agent integrations). It is recommended to enable no_proxy_nonexact_match to make the Agent match NO_PROXY values with the same rules (below) used for Agent integrations.

no_proxy_nonexact_match: true

The following rules apply to Agent integrations (and the whole Agent when no_proxy_nonexact_match is enabled):

  • A domain name matches that name and all subdomains, for example:
    • datadoghq.com matches app.agent.datadoghq.com, www.datadoghq.com, datadoghq.com, but not www.notdatadoghq.com
    • datadoghq matches frontend.datadoghq, backend.datadoghq, but not www.datadoghq.com nor www.datadoghq.eu
  • A domain name with a leading “.” matches subdomains only, for example:
    • .datadoghq.com matches app.agent.datadoghq.com, www.datadoghq.com, but not datadoghq.com
  • A CIDR range matches an IP address within the subnet, for example:
    • 192.168.1.0/24 matches IP range 192.168.1.1 through 192.168.1.254
  • An exact IP address, for example:
    • 169.254.169.254
  • A hostname, for example:
    • webserver1

Environment variables

Starting with Agent v6.4, you can set your proxy settings through environment variables:

  • DD_PROXY_HTTPS: Sets a proxy server for https requests.
  • DD_PROXY_HTTP: Sets a proxy server for http requests.
  • DD_PROXY_NO_PROXY: Sets a list of hosts that should bypass the proxy. The list is space-separated.

Environment variables have precedence over values in the datadog.yaml file. If the environment variables are present with an empty value, for example: DD_PROXY_HTTP="", the Agent uses the empty values instead of lower-precedence options.

On Unix hosts, a system-wide proxy might be specified using standard environment variables, such as HTTPS_PROXY, HTTP_PROXY, and NO_PROXY. The Agent uses these if present. Be careful, as such variables also impact every requests from integrations, including orchestrators like Docker, ECS, and Kubernetes.

The Agent uses the following values in order of precedence:

  1. DD_PROXY_HTTPS, DD_PROXY_HTTP, and DD_PROXY_NO_PROXY environment variables
  2. HTTPS_PROXY, HTTP_PROXY, and NO_PROXY environment variables
  3. Values inside datadog.yaml

Agent v5

The <HOST>:<PORT> used to proxy metrics can NOT be used to proxy logs. See the Proxy for Logs page.

Edit the datadog.conf file with your proxy information:

# If you need a proxy to connect to the Internet, provide the settings here
proxy_host: my-proxy.example.com
proxy_port: 3128
proxy_user: my_user
proxy_password: my_password

Do not forget to restart the Agent for the new settings to take effect.

Squid

Squid is a forward proxy for the web supporting HTTP, HTTPS, FTP, and more. It runs on most available operating systems, including Windows, and is licensed under the GNU GPL license. Squid is a straightforward option if you do not already have a running web proxy in your network.

Proxy forwarding with Squid

Configure Squid to only send traffic to Datadog

Install Squid on a host that has connectivity to both your internal Agents and Datadog. Use your operating system’s package manager, or install the software directly from Squid’s project page.

To configure Squid, edit the configuration file. This file is usually located at /etc/squid/squid.conf on Linux or C:\squid\etc\squid.conf in Windows.

Edit your squid.conf configuration file so that Squid is able to accept local traffic and forward it to the necessary Datadog intakes:

http_port 0.0.0.0:3128

acl local src 127.0.0.1/32

acl Datadog dstdomain .

http_access allow Datadog
http_access allow local manager
Start Squid

Start (or restart) Squid so that your new configurations can be applied.

sudo systemctl start squid

If Squid is already running, restart Squid instead with the following command:

sudo systemctl restart squid

If you are configuring Squid on Windows, you must first configure Squid as a system service. You can then run the following in an Administrator command prompt:

net start squid

If Squid is already running, restart Squid instead with the following commands:

net stop squid
net start squid
Datadog Agent configuration

Agent v6 & v7

Modify the Agent’s configuration file (datadog.yaml) to include the following:

proxy:
  http: http://127.0.0.1:3128
  https: http://127.0.0.1:3128

After saving these changes, restart the Agent.

Verify that Datadog is able to receive the data from your Agent(s) by checking your Infrastructure Overview.

Agent v5

Modify the Agent’s configuration file (datadog.conf) to include the following:

proxy_host: 127.0.0.1
proxy_port: 3128

After saving these changes, restart the Agent.

Verify that Datadog is able to receive the data from your Agent(s) by checking your Infrastructure Overview.

HAProxy

HAProxy is a free, fast, and reliable solution offering proxying for TCP and HTTP applications. While HAProxy is usually used as a load balancer to distribute incoming requests to pool servers, you can also use it to proxy Agent traffic to Datadog from hosts that have no outside connectivity:

agent ---> haproxy ---> Datadog

This is another good option if you do not have a web proxy readily available in your network and you wish to proxy a large number of Agents. In some cases, a single HAProxy instance is sufficient to handle local Agent traffic in your network, because each proxy can accommodate upwards of 1000 Agents.

Note: This figure is a conservative estimate based on the performance of m3.xl instances specifically. Numerous network-related and host-related variables can influence throughput of HAProxy, so you should keep an eye on your proxy deployment both before and after putting it into service. See the HAProxy documentation for additional information.

The communication between HAProxy and Datadog is always encrypted with TLS. The communication between the Agent host and the HAProxy host is not encrypted by default, because the proxy and the Agent are assumed to be on the same host. However, it is recommended that you secure this communication with TLS encryption if the HAproxy host and Agent host are not located on the same isolated local network. To encrypt data between the Agent and HAProxy, you need to create an x509 certificate with the Subject Alternative Name (SAN) extension for the HAProxy host. This certificate bundle (*.pem) should contain both the public certificate and private key. See this HAProxy blog post for more information.

Note: Download the Datadog certificate with one of the following commands:

sudo apt-get install ca-certificates # (Debian, Ubuntu)
yum install ca-certificates # (CentOS, Red Hat)

The path to the certificate is /etc/ssl/certs/ca-certificates.crt for Debian and Ubuntu, or /etc/ssl/certs/ca-bundle.crt for CentOS and Red Hat.

Proxy forwarding with HAProxy

HAProxy configuration

HAProxy should be installed on a host that has connectivity to Datadog. You can use one of the following configuration files if you do not already have it configured.

Note: It is recommended to use the HTTPS configuration file if the Agent and HAProxy are not part of the same isolated local network.

HTTP
# Basic configuration
global
    log 127.0.0.1 local0
    maxconn 4096
    stats socket /tmp/haproxy

# Some sane defaults
defaults
    log     global
    option  dontlognull
    retries 3
    option  redispatch
    timeout client 5s
    timeout server 5s
    timeout connect 5s

# This declares a view into HAProxy statistics, on port 3833
# You do not need credentials to view this page and you can
# turn it off once you are done with setup.
listen stats
    bind *:3833
    mode http
    stats enable
    stats uri /

# This section is to reload DNS Records
# Replace <DNS_SERVER_IP> and <DNS_SECONDARY_SERVER_IP> with your DNS Server IP addresses.
# For HAProxy 1.8 and newer
resolvers my-dns
    nameserver dns1 <DNS_SERVER_IP>:53
    nameserver dns2 <DNS_SECONDARY_SERVER_IP>:53
    resolve_retries 3
    timeout resolve 2s
    timeout retry 1s
    accepted_payload_size 8192
    hold valid 10s
    hold obsolete 60s

# This declares the endpoint where your Agents connects for
# sending metrics (for example, the value of "dd_url").
frontend metrics-forwarder
    bind *:3834
    mode http
    option tcplog
    default_backend datadog-metrics

    use_backend datadog-api if { path_beg -i  /api/v1/validate }
    use_backend datadog-flare if { path_beg -i  /support/flare/ }

# This declares the endpoint where your Agents connects for
# sending traces (for example, the value of "endpoint" in the APM
# configuration section).
frontend traces-forwarder
    bind *:3835
    mode tcp
    option tcplog
    default_backend datadog-traces

# This declares the endpoint where your Agents connects for
# sending profiles (for example, the value of "apm_config.profiling_dd_url").
frontend profiles-forwarder
    bind *:3836
    mode tcp
    option tcplog
    default_backend datadog-profiles

# This declares the endpoint where your agents connects for
# sending processes (for example, the value of "url" in the process
# configuration section).
frontend processes-forwarder
    bind *:3837
    mode tcp
    option tcplog
    default_backend datadog-processes

# This declares the endpoint where your Agents connects for
# sending Logs (e.g the value of "logs.config.logs_dd_url")
# If sending logs with use_http: true
frontend logs_http_frontend
    bind *:3838
    mode http
    option tcplog
    default_backend datadog-logs-http

# If sending logs with use_tcp: true
# frontend logs_frontend
#    bind *:10514
#    mode tcp
#    option tcplog
#    default_backend datadog-logs

# This declares the endpoint where your Agents connects for
# sending database monitoring metrics and activity (e.g the value of "database_monitoring.metrics.dd_url" and "database_monitoring.activity.dd_url")
frontend database_monitoring_metrics_frontend
    bind *:3839
    mode http
    option tcplog
    default_backend datadog-database-monitoring-metrics

# This declares the endpoint where your Agents connects for
# sending database monitoring samples (e.g the value of "database_monitoring.samples.dd_url")
frontend database_monitoring_samples_frontend
    bind *:3840
    mode http
    option tcplog
    default_backend datadog-database-monitoring-samples

# This declares the endpoint where your Agents connects for
# sending Network Devices Monitoring metadata (e.g the value of "network_devices.metadata.dd_url")
frontend network_devices_metadata_frontend
    bind *:3841
    mode http
    option tcplog
    default_backend datadog-network-devices-metadata

# This declares the endpoint where your Agents connects for
# sending Network Devices SNMP Traps data (e.g the value of "network_devices.snmp_traps.forwarder.dd_url")
frontend network_devices_snmp_traps_frontend
    bind *:3842
    mode http
    option tcplog
    default_backend datadog-network-devices-snmp-traps

# This declares the endpoint where your Agents connect for
# sending Instrumentation Telemetry data (e.g. the value of "apm_config.telemetry.dd_url")
frontend instrumentation_telemetry_data_frontend
    bind *:3843
    mode tcp
    option tcplog
    default_backend datadog-instrumentations-telemetry

# This declares the endpoint where your Agents connects for
# sending Network Devices Monitoring NetFlow flows (for example, the value of "network_devices.netflow.dd_url")
frontend network_devices_netflow_frontend
    bind *:3845
    mode http
    option tcplog
    default_backend datadog-network-devices-netflow

# This is the Datadog server. In effect, any TCP request coming
# to the forwarder frontends defined above are proxied to
# Datadog's public endpoints.
backend datadog-metrics
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 haproxy-app.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership haproxy-app.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-api
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 api.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership api.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-flare
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 flare.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership flare.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-traces
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 trace.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership trace.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-profiles
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 intake.profile.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership profile.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-processes
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 process.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership process.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-logs-http
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server datadog agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-database-monitoring-metrics
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 dbm-metrics-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server datadog agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-database-monitoring-samples
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 dbquery-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server datadog agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-network-devices-metadata
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 ndm-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership ndm-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-network-devices-snmp-traps
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 snmp-traps-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership snmp-traps-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-instrumentations-telemetry
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 instrumentation-telemetry-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership instrumentation-telemetry-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-network-devices-netflow
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 ndmflow-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership ndmflow-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>
HTTPS

This configuration adds SSL/TLS encryption on communication between the Agent and HAProxy. Replace the variable <PATH_TO_PROXY_CERTIFICATE_PEM> with the path to the proxy certificate bundle (*.pem).

# Basic configuration
global
    log 127.0.0.1 local0
    maxconn 4096
    stats socket /tmp/haproxy

# Some sane defaults
defaults
    log     global
    option  dontlognull
    retries 3
    option  redispatch
    timeout client 5s
    timeout server 5s
    timeout connect 5s

# This declares a view into HAProxy statistics, on port 3833
# You do not need credentials to view this page and you can
# turn it off once you are done with setup.
listen stats
    bind *:3833
    mode http
    stats enable
    stats uri /

# This section is to reload DNS Records
# Replace <DNS_SERVER_IP> and <DNS_SECONDARY_SERVER_IP> with your DNS Server IP addresses.
# For HAProxy 1.8 and newer
resolvers my-dns
    nameserver dns1 <DNS_SERVER_IP>:53
    nameserver dns2 <DNS_SECONDARY_SERVER_IP>:53
    resolve_retries 3
    timeout resolve 2s
    timeout retry 1s
    accepted_payload_size 8192
    hold valid 10s
    hold obsolete 60s

# This declares the endpoint where your Agents connect for
# sending metrics (for example, the value of "dd_url").
frontend metrics-forwarder
    bind *:3834 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-metrics

    use_backend datadog-api if { path_beg -i  /api/v1/validate }
    use_backend datadog-flare if { path_beg -i  /support/flare/ }

# This declares the endpoint where your Agents connect for
# sending traces (for example, the value of "endpoint" in the APM
# configuration section).
frontend traces-forwarder
    bind *:3835 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode tcp
    option tcplog
    default_backend datadog-traces

# This declares the endpoint where your Agents connect for
# sending profiles (for example, the value of "apm_config.profiling_dd_url").
frontend profiles-forwarder
    bind *:3836 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode tcp
    option tcplog
    default_backend datadog-profiles

# This declares the endpoint where your Agents connect for
# sending processes (for example, the value of "url" in the process
# configuration section).
frontend processes-forwarder
    bind *:3837 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode tcp
    option tcplog
    default_backend datadog-processes

# This declares the endpoint where your Agents connect for
# sending Logs (e.g the value of "logs.config.logs_dd_url")
# If sending logs with use_http: true
frontend logs_http_frontend
    bind *:3838 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-logs-http

# If sending logs with use_tcp: true
# frontend logs_frontend
#    bind *:10514 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
#    mode tcp
#    option tcplog
#    default_backend datadog-logs

# This declares the endpoint where your Agents connect for
# sending database monitoring metrics and activity (e.g the value of "database_monitoring.metrics.dd_url" and "database_monitoring.activity.dd_url")
frontend database_monitoring_metrics_frontend
    bind *:3839 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-database-monitoring-metrics

# This declares the endpoint where your Agents connect for
# sending database monitoring samples (e.g the value of "database_monitoring.samples.dd_url")
frontend database_monitoring_samples_frontend
    bind *:3840 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-database-monitoring-samples

# This declares the endpoint where your Agents connect for
# sending Network Devices Monitoring metadata (e.g the value of "network_devices.metadata.dd_url")
frontend network_devices_metadata_frontend
    bind *:3841 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-network-devices-metadata

# This declares the endpoint where your Agents connect for
# sending Network Devices SNMP Traps data (e.g the value of "network_devices.snmp_traps.forwarder.dd_url")
frontend network_devices_snmp_traps_frontend
    bind *:3842 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-network-devices-snmp-traps


# This declares the endpoint where your Agents connect for
# sending Instrumentation Telemetry data (e.g. the value of "apm_config.telemetry.dd_url")
frontend instrumentation_telemetry_data_frontend
    bind *:3843 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode tcp
    option tcplog
    default_backend datadog-instrumentations-telemetry

# This declares the endpoint where your Agents connect for
# sending Network Devices Monitoring NetFlow flows (for example, the value of "network_devices.netflow.dd_url")
frontend network_devices_netflow_frontend
    bind *:3845 ssl crt <PATH_TO_PROXY_CERTIFICATE_PEM>
    mode http
    option tcplog
    default_backend datadog-network-devices-netflow

# This is the Datadog server. In effect any TCP request coming
# to the forwarder frontends defined above are proxied to
# Datadog's public endpoints.
backend datadog-metrics
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 haproxy-app.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership haproxy-app.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-api
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 api.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership api.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-flare
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 flare.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership flare.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-traces
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 trace.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership trace.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-profiles
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 intake.profile.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership profile.agent.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-processes
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 process.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership process.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-logs-http
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 agent-http-intake.logs.:443  check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server datadog agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-database-monitoring-metrics
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 dbm-metrics-intake.:443  check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server datadog agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-database-monitoring-samples
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 dbquery-intake.:443  check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server datadog agent-http-intake.logs.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-network-devices-metadata
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 ndm-intake.:443  check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership ndm-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-network-devices-snmp-traps
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 snmp-traps-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership snmp-traps-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-instrumentations-telemetry
    balance roundrobin
    mode tcp
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 instrumentation-telemetry-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership instrumentation-telemetry-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

backend datadog-network-devices-netflow
    balance roundrobin
    mode http
    # The following configuration is for HAProxy 1.8 and newer
    server-template mothership 5 ndmflow-intake.:443  check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES> check resolvers my-dns init-addr none resolve-prefer ipv4
    # Uncomment the following configuration for older HAProxy versions
    # server mothership ndmflow-intake.:443 check port 443 ssl verify required ca-file <PATH_TO_CERTIFICATES>

Note: You can use verify none instead of verify required ca-file <PATH_TO_CERTIFICATES> if you are unable to get the certificates on the proxy host, but be aware that HAProxy will not be able to verify Datadog’s intake certificate in that case.

HAProxy 1.8 and newer allow DNS service discovery to detect server changes and automatically apply them to your configuration. If you are using older version of HAProxy, you have to reload or restart HAProxy. It is recommended to have a cron job reload HAProxy every 10 minutes (such as service haproxy reload) to force a refresh of HAProxy’s DNS cache, in case fails over to another IP.

Datadog Agent configuration

Agent v6 & v7

Edit each Agent to point to HAProxy by setting its dd_url to the address of HAProxy, for example: haproxy.example.com. This dd_url setting can be found in the datadog.yaml file.

dd_url: <SCHEME>://haproxy.example.com:3834

Replace <SCHEME> with https if you previously chose the HAProxy HTTPS configuration, or with http if you did not choose HTTPS.

To send traces, profiles, processes, and logs through the proxy, setup the following in the datadog.yaml file:

apm_config:
    apm_dd_url: <SCHEME>://haproxy.example.com:3835
    profiling_dd_url: <SCHEME>://haproxy.example.com:3836
    telemetry:
        dd_url: <SCHEME>://haproxy.example.com:3843

process_config:
    process_dd_url: <SCHEME>://haproxy.example.com:3837

logs_config:
    use_http: true
    logs_dd_url: haproxy.example.com:3838
    # Comment the line below to use encryption between the Agent and HAProxy
    logs_no_ssl: true

database_monitoring:
    metrics:
        logs_dd_url: haproxy.example.com:3839
        # Comment the line below to use encryption between the Agent and HAProxy
        logs_no_ssl: true
    activity:
        logs_dd_url: haproxy.example.com:3839
        # Comment the line below to use encryption between the Agent and HAProxy
        logs_no_ssl: true
    samples:
        logs_dd_url: haproxy.example.com:3840
        # Comment the line below to use encryption between the Agent and HAProxy
        logs_no_ssl: true

network_devices:
    metadata:
        logs_dd_url: haproxy.example.com:3841
        # Comment the line below to use encryption between the Agent and HAProxy
        logs_no_ssl: true
    snmp_traps:
        forwarder:
            logs_dd_url: haproxy.example.com:3842
            # Comment the line below to use encryption between the Agent and HAProxy
            logs_no_ssl: true
    netflow:
        forwarder:
            logs_dd_url: haproxy.example.com:3845
            # Comment the line below to use encryption between the Agent and HAProxy
            logs_no_ssl: true

When using encryption between the Agent and HAProxy, if the Agent does not have access to the proxy certificate, is unable to validate it, or the validation is not needed, you can edit the datadog.yaml Agent configuration file and set skip_ssl_validation to true. With this option set to true, the Agent skips the certificate validation step and does not verify the identity of the proxy, but the communication is still encrypted with SSL/TLS.

skip_ssl_validation: true

Finally restart the Agent.

To verify that everything is working properly, review the HAProxy statistics at http://haproxy.example.com:3833 as well as the Infrastructure Overview.

Agent v5

Edit each Agent to point to HAProxy by setting its dd_url to the address of HAProxy, for example: haproxy.example.com. This dd_url setting can be found in the datadog.conf file.

dd_url: http://haproxy.example.com:3834

To send traces or processes through the proxy, setup the following in the datadog.conf file:

[trace.api]
endpoint = http://haproxy.example.com:3835

[process.api]
endpoint = http://haproxy.example.com:3837

Edit your supervisor configuration to disable SSL certificate verification. This is needed to prevent Python from complaining about the discrepancy between the hostname on the SSL certificate (app.datadoghq.com) and your HAProxy hostname. The supervisor configuration found at:

  • /etc/dd-agent/supervisor_ddagent.conf on Debian-based systems
  • /etc/dd-agent/supervisor.conf on Red Hat-based systems
  • /opt/local/datadog/supervisord/supervisord.conf on SmartOS
  • /usr/local/etc/datadog/supervisord/supervisord.conf on FreeBSD
  • ~/.datadog-agent/supervisord/supervisord.conf on macOS

Assuming that the supervisor file is found at <SUP_FILE>

sed -i 's/ddagent.py/ddagent.py --sslcheck=0/' <SUP_FILE>

For the Windows Agent, edit your configuration file datadog.conf and add this option:

skip_ssl_validation: yes

Finally restart the Agent.

To verify that everything is working properly, review the HAProxy statistics at http://haproxy.example.com:3833 as well as the Infrastructure Overview.

NGINX

NGINX is a web server which can also be used as a reverse proxy, load balancer, mail proxy, and HTTP cache. You can also use NGINX as a proxy for your Datadog Agents:

agent ---> nginx ---> Datadog

The communication between NGINX and Datadog is always encrypted with TLS. The communication between the Agent host and the NGINX host is not encrypted by default, because the proxy and the Agent are assumed to be on the same host. However, it is recommended that you secure this communication with TLS encryption if they are not located on the same isolated local network. In order to encrypt data between the Agent and NGINX, you need to create an x509 certificate with the Subject Alternative Name (SAN) extension for the NGINX host.

Note: Download the Datadog certificate with one of the following commands:

sudo apt-get install ca-certificates # (Debian, Ubuntu)
yum install ca-certificates # (CentOS, Red Hat)

The path to the certificate is /etc/ssl/certs/ca-certificates.crt for Debian and Ubuntu or /etc/ssl/certs/ca-bundle.crt for CentOS and Red Hat.

Proxy forwarding with NGINX

NGINX configuration

NGINX should be installed on a host that has connectivity to Datadog. You can use one of the following configuration files if you do not already have it configured.

Note: It is recommended to use the HTTPS configuration file if the Agent and NGINX are not part of the same isolated local network.

HTTP
user nginx;
worker_processes auto;
error_log /var/log/nginx/error.log;
pid /run/nginx.pid;

events {
    worker_connections 1024;
}
# HTTP Proxy for Datadog Agent
http {

    proxy_ssl_trusted_certificate <PATH_TO_CERTIFICATES>;

    server {
        listen 3834; #listen for metrics
        access_log off;

        location /api/v1/validate {
            proxy_ssl_verify on;
            proxy_pass https://api.:443/api/v1/validate;
        }
        location /support/flare/ {
            proxy_ssl_verify on;
            proxy_pass https://flare.:443/support/flare/;
        }
        location / {
            proxy_ssl_verify on;
            proxy_pass https://haproxy-app.agent.:443/;
        }
    }
}
# TCP Proxy for Datadog Agent
stream {

    proxy_ssl_trusted_certificate <PATH_TO_CERTIFICATES>;

    server {
        listen 3835; #listen for traces
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass trace.agent.:443;
    }
    server {
        listen 3836; #listen for profiles
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass intake.profile.:443;
    }
    server {
        listen 3837; #listen for processes
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass process.:443;
    }
    server {
        listen 3838; #listen for logs with use_http: true
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass agent-http-intake.logs.:443;
    }
    server {
        listen 3839; #listen for database monitoring metrics
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass dbm-metrics-intake.:443;
    }
    server {
        listen 3840; #listen for database monitoring samples
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass dbquery-intake.:443;
    }
    server {
        listen 3841; #listen for network devices metadata
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass ndm-intake.:443;
    }
    server {
        listen 3842; #listen for network devices traps
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass snmp-traps-intake.:443;
    }
    server {
        listen 3843; #listen for instrumentations telemetry data
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass instrumentation-telemetry-intake.:443;
    }
    server {
        listen 3845; #listen for network devices netflow
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass ndmflow-intake.:443;
    }
}
HTTPS

This configuration adds SSL/TLS encryption on communication between the Agent and NGINX. Replace <PATH_TO_PROXY_CERTIFICATE> with the path to the proxy public certificate and <PATH_TO_PROXY_CERTIFICATE_KEY> with the path to the private key.

user nginx;
worker_processes auto;
error_log /var/log/nginx/error.log;
pid /run/nginx.pid;

events {
    worker_connections 1024;
}
# HTTP Proxy for Datadog Agent
http {

    proxy_ssl_trusted_certificate <PATH_TO_CERTIFICATES>;

    ssl_certificate     <PATH_TO_PROXY_CERTIFICATE>;
    ssl_certificate_key <PATH_TO_PROXY_CERTIFICATE_KEY>;

    server {
        listen 3834 ssl; #listen for metrics
        access_log off;

        location /api/v1/validate {
            proxy_ssl_verify on;
            proxy_pass https://api.:443/api/v1/validate;
        }
        location /support/flare/ {
            proxy_ssl_verify on;
            proxy_pass https://flare.:443/support/flare/;
        }
        location / {
            proxy_ssl_verify on;
            proxy_pass https://haproxy-app.agent.:443/;
        }
    }
}
# TCP Proxy for Datadog Agent
stream {

    proxy_ssl_trusted_certificate <PATH_TO_CERTIFICATES>;

    ssl_certificate     <PATH_TO_PROXY_CERTIFICATE>;
    ssl_certificate_key <PATH_TO_PROXY_CERTIFICATE_KEY>;

    server {
        listen 3835 ssl; #listen for traces
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass trace.agent.:443;
    }
    server {
        listen 3836 ssl; #listen for profiles
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass intake.profile.:443;
    }
    server {
        listen 3837 ssl; #listen for processes
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass process.:443;
    }
    server {
        listen 3838 ssl; #listen for logs with use_http: true
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass agent-http-intake.logs.:443;
    }
    server {
        listen 3839 ssl; #listen for database monitoring metrics
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass dbm-metrics-intake.:443;
    }
    server {
        listen 3840 ssl; #listen for database monitoring samples
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass dbquery-intake.:443;
    }
    server {
        listen 3841 ssl; #listen for network devices metadata
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass ndm-intake.:443;
    }
    server {
        listen 3842 ssl; #listen for network devices traps
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass snmp-traps-intake.:443;
    }
    server {
        listen 3843 ssl; #listen for instrumentations telemetry data
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass instrumentation-telemetry-intake.:443;
    }
    server {
        listen 3845 ssl; #listen for network devices netflow
        proxy_ssl_verify on;
        proxy_ssl on;
        proxy_pass ndmflow-intake.:443;
    }
}

Note: You can remove proxy_ssl_verify on if you are unable to get the certificates on the proxy host, but be aware that NGINX will not be able to verify Datadog’s intake certificate in that case.

Datadog Agent configuration

Edit each Agent configuration file to point to Nginx by setting its dd_url to the address of Nginx, for example: nginx.example.com. This dd_url setting can be found in the datadog.yaml file.

dd_url: "<SCHEME>://nginx.example.com:3834"

Replace <SCHEME> with https if you previously chose the HAProxy HTTPS configuration, or with http if you did not choose HTTPS.

To send traces, profiles, processes, and logs through the proxy, setup the following in the datadog.yaml file:

apm_config:
    apm_dd_url: <SCHEME>://nginx.example.com:3835
    profiling_dd_url: <SCHEME>://nginx.example.com:3836
    telemetry:
        dd_url: <SCHEME>://nginx.example.com:3843

process_config:
    process_dd_url: <SCHEME>://nginx.example.com:3837

logs_config:
    use_http: true
    logs_dd_url: nginx.example.com:3838
    # Comment the line below to use encryption between the Agent and NGINX
    logs_no_ssl: true

database_monitoring:
    metrics:
        logs_dd_url: nginx.example.com:3839
        # Comment the line below to use encryption between the Agent and NGINX
        logs_no_ssl: true
    activity:
        logs_dd_url: nginx.example.com:3839
        # Comment the line below to use encryption between the Agent and NGINX
        logs_no_ssl: true
    samples:
        logs_dd_url: nginx.example.com:3840
        # Comment the line below to use encryption between the Agent and NGINX
        logs_no_ssl: true

network_devices:
    metadata:
        logs_dd_url: nginx.example.com:3841
        # Comment the line below to use encryption between the Agent and NGINX
        logs_no_ssl: true
    snmp_traps:
        forwarder:
            logs_dd_url: nginx.example.com:3842
            # Comment the line below to use encryption between the Agent and NGINX
            logs_no_ssl: true
    netflow:
        forwarder:
            logs_dd_url: nginx.example.com:3845
            # Comment the line below to use encryption between the Agent and NGINX
            logs_no_ssl: true

When using encryption between the Agent and NGINX, if the Agent does not have access to the proxy certificate, is unable to validate it, or the validation is not needed, you can edit the datadog.yaml Agent configuration file and set skip_ssl_validation to true. With this option set to true, the Agent skips the certificate validation step and does not verify the identity of the proxy, but the communication is still encrypted with SSL/TLS.

skip_ssl_validation: true

When sending logs over TCP, see TCP Proxy for Logs.

Datadog Agent

This feature is only available for Agent v5.

It is recommended to use an actual proxy (a web proxy or HAProxy) to forward your traffic to Datadog, however if those options aren’t available to you, it is possible to configure an instance of Agent v5 to serve as a proxy.

  1. Designate one node running datadog-agent as the proxy. In this example assume that the proxy name is proxy-node. This node must be able to reach https://app.datadoghq.com.

  2. Verify SSL connectivity on proxy-node:

    curl -v https://app.datadoghq.com/account/login 2>&1 | grep "200 OK"
    
  3. Allow non-local traffic on proxy-node by changing the following line in datadog.conf. # non_local_traffic: no should read non_local_traffic: yes.

  4. Make sure proxy-node can be reached from the other nodes over port 17123. Start the Agent on the proxy-node and run on the other nodes:

    curl -v http://proxy-node:17123/status 2>&1 | grep "200 OK"

  5. Update non-proxy nodes to forward to proxy-node. Change the following line in datadog.conf from:

    dd_url: https://app.datadoghq.com to dd_url: http://proxy-node:17123

  6. Verify on the Infrastructure page that all nodes report data to Datadog.

Further Reading