Skip to content

Instantly share code, notes, and snippets.

@poolski
Last active February 8, 2023 03:28
Show Gist options
  • Save poolski/9318b70285379d884422b2419c0325c9 to your computer and use it in GitHub Desktop.
Save poolski/9318b70285379d884422b2419c0325c9 to your computer and use it in GitHub Desktop.
Sample Envoy Config
---
# Envoy requires this to be configured, even if you don't intend to use the admin interface.
admin:
access_log_path: "/var/log/envoy/envoy-admin.log"
address:
socket_address:
address: 127.0.0.1
port_value: 8001
# Identify your node/pod/instance/whatever.
node:
cluster: production
id: my-service-node-1
# Set up stats sink(s). This example uses statsD but more can be found here:
# https://www.envoyproxy.io/docs/envoy/v1.9.0/api-v2/config/metrics/v2/stats.proto
stats_sinks:
- name: envoy.statsd
config:
address:
socket_address:
address: 127.0.0.1
port_value: 8125
stats_config:
stats_matcher:
exclusion_list:
# A sample list of values that we're not shipping to the statsD sink because we don't need them.
# https://www.envoyproxy.io/docs/envoy/v1.9.0/api-v2/type/matcher/string.proto#envoy-api-msg-type-matcher-liststringmatcher
patterns:
- regex: cluster.([^.]*).update_attempt$
- regex: cluster.([^.]*).update_no_rebuild$
- regex: cluster.([^.]*).max_host_weight$
- regex: cluster.([^.]*).health_check.attempt
- regex: cluster.([^.]*).membership_total
static_resources:
listeners:
- name: internal-grpc-traffic
address:
socket_address:
# The address and port you want Envoy to listen on.
# You can specify multiple listeners, each with their own filter chains.
address: 127.0.0.1
port_value: 8888
filter_chains:
- filters:
# Create an instance of the HTTP connection manager built-in filter. More here:
# https://www.envoyproxy.io/docs/envoy/v1.9.0/api-v2/api/v2/listener/listener.proto#envoy-api-msg-listener-filter
- name: envoy.http_connection_manager
config:
access_log:
- name: envoy.file_access_log
config:
path: "/var/log/envoy/access.log"
# A slightly better-structure log format, using JSON rather than the default combined log format.
# This allows us to pipe the log output through something like JQ to make log events more readable
# See https://www.envoyproxy.io/docs/envoy/v1.9.0/configuration/access_log#config-access-log
# for more fomatting options.
json_format:
start_time: "%START_TIME%"
request_id: "%REQ(REQUEST-ID)%"
caller: "%REQ(CALLER)%"
request_method: "%REQ(:METHOD)%"
request_path: "%REQ(X-ENVOY-ORIGINAL-PATH?:PATH)%"
content_type: "%REQ(CONTENT-TYPE)%"
protocol: "%PROTOCOL%"
response_code: "%RESPONSE_CODE%"
response_flags: "%RESPONSE_FLAGS%"
bytes_sent: "%BYTES_SENT%"
bytes_received: "%BYTES_RECEIVED%"
request_duration: "%DURATION%"
response_duration: "%RESPONSE_DURATION%"
upstream_response_time: "%RESP(X-ENVOY-UPSTREAM-SERVICE-TIME)%"
client_address: "%DOWNSTREAM_REMOTE_ADDRESS_WITHOUT_PORT%"
x_forwarded_for: "%REQ(X-FORWARDED-FOR)%"
user_agent: "%REQ(USER-AGENT)%"
http2_authority: "%REQ(:AUTHORITY)%"
upstream_cluster: "%UPSTREAM_CLUSTER%"
upstream_host: "%UPSTREAM_HOST%"
stat_prefix: ingress_grpc
codec_type: AUTO
route_config:
name: local_route
# Create your virtual hosts here. You can specify multiple domains to match on
# or you can use path-based routing if that's easier. More detailed docs here:
# https://www.envoyproxy.io/docs/envoy/v1.9.0/api-v2/api/v2/route/route.proto#route-virtualhost
virtual_hosts:
- name: awesome-service
# Which Host or :authority headers are you matching?
domains:
- awesome-service.envoy.example.com
routes:
# You can match on the root of the domain or get really creative:
# https://www.envoyproxy.io/docs/envoy/v1.9.0/api-v2/api/v2/route/route.proto#envoy-api-msg-route-routematch
- match:
prefix: "/"
grpc: {}
# Where should matching requests be sent?
# https://www.envoyproxy.io/docs/envoy/v1.9.0/api-v2/api/v2/route/route.proto#envoy-api-msg-route-routeaction
route:
cluster: awesome-service-grpc
timeout:
seconds: 15
# Use the built-in 'router' resource
http_filters:
- name: envoy.router
# Set up your TLS certificates here.
tls_context:
common_tls_context:
alpn_protocols: h2
tls_certificates:
- certificate_chain:
filename: "/etc/envoy/ssl/wildcard.envoy.crt"
private_key:
filename: "/etc/envoy/ssl/wildcard.envoy.key"
clusters:
# Set up your cluster parameters here.
- name: awesome-service-grpc
connect_timeout: 0.5s
http2_protocol_options: {}
type: strict_dns # Strict DNS allows for DNS lookups rather than hard-coding IPs
lb_policy: round_robin
health_checks: # Not absolutely necessary, but useful to explicitly configure.
- grpc_health_check:
authority: awesome-service.internal-domain.com # Custom HTTP/2 :authority header
timeout: 1s
interval: 2s
interval_jitter: 1s
healthy_threshold: 3
unhealthy_threshold: 3
event_log_path: "/var/log/envoy/healthcheck.log"
tls_context: # What Host header should **Envoy** send to the upstream service?
sni: awesome-service.internal-domain.com
hosts:
- socket_address:
address: awesome-service.service.consul # Where is the gRPC service located? Let's ask Consul.
port_value: 5100
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment