tests: adding new envoy integration tests for L7 service-resolvers (#6129)

Additionally:

- wait for bootstrap config entries to be applied

- run the verify container in the host's PID namespace so we can kill
envoys without mounting the docker socket

* assert that we actually send HEALTHY and UNHEALTHY endpoints down in EDS during failover

test/integration/connect/envoy/case-badauthz/setup.sh

@@ -8,4 +8,4 @@ docker_consul intention create -deny s1 s2
 gen_envoy_bootstrap s1 19000
 gen_envoy_bootstrap s2 19001
 
-export REQUIRED_SERVICES="s1 s1-sidecar-proxy s2 s2-sidecar-proxy"
+export REQUIRED_SERVICES="s1 s1-sidecar-proxy s2 s2-sidecar-proxy"

test/integration/connect/envoy/case-badauthz/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should NOT be able to connect to s2" {

test/integration/connect/envoy/case-basic/verify.bats

@@ -27,7 +27,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2" {

test/integration/connect/envoy/case-centralconf/setup.sh

@@ -2,8 +2,12 @@
 
 set -euo pipefail
 
-# retry because resolving the central config might race
-retry_default gen_envoy_bootstrap s1 19000
-retry_default gen_envoy_bootstrap s2 19001
+# wait for bootstrap to apply config entries
+wait_for_config_entry proxy-defaults global
+wait_for_config_entry service-defaults s1
+wait_for_config_entry service-defaults s2
 
-export REQUIRED_SERVICES="s1 s1-sidecar-proxy s2 s2-sidecar-proxy"
+gen_envoy_bootstrap s1 19000
+gen_envoy_bootstrap s2 19001
+
+export REQUIRED_SERVICES="s1 s1-sidecar-proxy s2 s2-sidecar-proxy"

test/integration/connect/envoy/case-centralconf/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2 with http/1.1" {

test/integration/connect/envoy/case-cfg-resolver-defaultsubset/config_entries.hcl

@@ -0,0 +1,28 @@
+enable_central_service_config = true
+
+config_entries {
+  bootstrap {
+    kind = "proxy-defaults"
+    name = "global"
+
+    config {
+      protocol = "http"
+    }
+  }
+
+  bootstrap {
+    kind           = "service-resolver"
+    name           = "s2"
+    default_subset = "v2"
+
+    subsets = {
+      "v1" = {
+        filter = "Service.Meta.version == v1"
+      }
+
+      "v2" = {
+        filter = "Service.Meta.version == v2"
+      }
+    }
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-defaultsubset/s2-v1.hcl

@@ -0,0 +1,13 @@
+services {
+  id   = "s2-v1"
+  name = "s2"
+  port = 8182
+
+  meta {
+    version = "v1"
+  }
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-defaultsubset/s2-v2.hcl

@@ -0,0 +1,13 @@
+services {
+  id   = "s2-v2"
+  name = "s2"
+  port = 8183
+
+  meta {
+    version = "v2"
+  }
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-defaultsubset/setup.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+
+set -euo pipefail
+
+# wait for bootstrap to apply config entries
+wait_for_config_entry proxy-defaults global
+wait_for_config_entry service-resolver s2
+
+# s2 is retained just to have a honeypot for bad envoy configs to route into
+gen_envoy_bootstrap s1 19000
+gen_envoy_bootstrap s2-v1 19001
+gen_envoy_bootstrap s2-v2 19002
+gen_envoy_bootstrap s2 19003
+
+export REQUIRED_SERVICES="
+s1 s1-sidecar-proxy
+s2 s2-sidecar-proxy
+s2-v1 s2-v1-sidecar-proxy
+s2-v2 s2-v2-sidecar-proxy
+"

test/integration/connect/envoy/case-cfg-resolver-defaultsubset/verify.bats

@@ -0,0 +1,48 @@
+#!/usr/bin/env bats
+
+load helpers
+
+@test "s1 proxy admin is up on :19000" {
+  retry_default curl -f -s localhost:19000/stats -o /dev/null
+}
+
+@test "s2-v1 proxy admin is up on :19001" {
+  retry_default curl -f -s localhost:19001/stats -o /dev/null
+}
+
+@test "s2-v2 proxy admin is up on :19002" {
+  retry_default curl -f -s localhost:19002/stats -o /dev/null
+}
+
+@test "s2 proxy admin is up on :19003" {
+  retry_default curl -f -s localhost:19003/stats -o /dev/null
+}
+
+@test "s1 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21000 s1
+}
+
+@test "s2-v1 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21001 s2
+}
+
+@test "s2-v2 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21002 s2
+}
+
+@test "s2 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21003 s2
+}
+
+@test "s2 proxies should be healthy" {
+  assert_service_has_healthy_instances s2 3
+}
+
+@test "s1 upstream should have healthy endpoints for v2.s2" {
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 v2.s2 HEALTHY 1
+}
+
+@test "s1 upstream should be able to connect to s2-v2 via upstream s2" {
+  assert_expected_fortio_name s2-v2
+}
+

test/integration/connect/envoy/case-cfg-resolver-subset-redirect/config_entries.hcl

@@ -0,0 +1,35 @@
+config_entries {
+  bootstrap {
+    kind = "proxy-defaults"
+    name = "global"
+
+    config {
+      protocol = "http"
+    }
+  }
+
+  bootstrap {
+    kind = "service-resolver"
+    name = "s3"
+
+    subsets = {
+      "v1" = {
+        filter = "Service.Meta.version == v1"
+      }
+
+      "v2" = {
+        filter = "Service.Meta.version == v2"
+      }
+    }
+  }
+
+  bootstrap {
+    kind = "service-resolver"
+    name = "s2"
+
+    redirect {
+      service        = "s3"
+      service_subset = "v2"
+    }
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-subset-redirect/s3-v1.hcl

@@ -0,0 +1,13 @@
+services {
+  id   = "s3-v1"
+  name = "s3"
+  port = 8283
+
+  meta {
+    version = "v1"
+  }
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-subset-redirect/s3-v2.hcl

@@ -0,0 +1,13 @@
+services {
+  id   = "s3-v2"
+  name = "s3"
+  port = 8284
+
+  meta {
+    version = "v2"
+  }
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-subset-redirect/s3.hcl

@@ -0,0 +1,8 @@
+services {
+  name = "s3"
+  port = 8282
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-subset-redirect/setup.sh

@@ -0,0 +1,23 @@
+#!/bin/bash
+
+set -euo pipefail
+
+# wait for bootstrap to apply config entries
+wait_for_config_entry proxy-defaults global
+wait_for_config_entry service-resolver s2
+wait_for_config_entry service-resolver s3
+
+# s2, s3, and s3-v1 are retained just to have a honeypot for bad envoy configs to route into
+gen_envoy_bootstrap s1 19000
+gen_envoy_bootstrap s2 19001
+gen_envoy_bootstrap s3-v1 19002
+gen_envoy_bootstrap s3-v2 19003
+gen_envoy_bootstrap s3 19004
+
+export REQUIRED_SERVICES="
+s1 s1-sidecar-proxy
+s2 s2-sidecar-proxy
+s3 s3-sidecar-proxy
+s3-v1 s3-v1-sidecar-proxy
+s3-v2 s3-v2-sidecar-proxy
+"

test/integration/connect/envoy/case-cfg-resolver-subset-redirect/verify.bats

@@ -0,0 +1,56 @@
+#!/usr/bin/env bats
+
+load helpers
+
+@test "s1 proxy admin is up on :19000" {
+  retry_default curl -f -s localhost:19000/stats -o /dev/null
+}
+
+@test "s2 proxy admin is up on :19001" {
+  retry_default curl -f -s localhost:19001/stats -o /dev/null
+}
+
+@test "s3-v1 proxy admin is up on :19002" {
+  retry_default curl -f -s localhost:19002/stats -o /dev/null
+}
+
+@test "s3-v2 proxy admin is up on :19003" {
+  retry_default curl -f -s localhost:19003/stats -o /dev/null
+}
+
+@test "s3 proxy admin is up on :19004" {
+  retry_default curl -f -s localhost:19004/stats -o /dev/null
+}
+
+@test "s1 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21000 s1
+}
+
+@test "s2 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21001 s2
+}
+
+@test "s3-v1 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21002 s3
+}
+
+@test "s3-v2 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21002 s3
+}
+
+@test "s3 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21003 s3
+}
+
+@test "s3 proxies should be healthy" {
+  assert_service_has_healthy_instances s3 3
+}
+
+@test "s1 upstream should have healthy endpoints for v2.s3" {
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 v2.s3 HEALTHY 1
+}
+
+@test "s1 upstream should be able to connect to s3-v2 via upstream s2" {
+  assert_expected_fortio_name s3-v2
+}
+

test/integration/connect/envoy/case-cfg-resolver-svc-failover/config_entries.hcl

@@ -0,0 +1,39 @@
+enable_central_service_config = true
+
+config_entries {
+  bootstrap {
+    kind = "proxy-defaults"
+    name = "global"
+
+    config {
+      protocol = "http"
+    }
+  }
+
+  bootstrap {
+    kind = "service-resolver"
+    name = "s3"
+
+    subsets = {
+      "v1" = {
+        filter = "Service.Meta.version == v1"
+      }
+
+      "v2" = {
+        filter = "Service.Meta.version == v2"
+      }
+    }
+  }
+
+  bootstrap {
+    kind = "service-resolver"
+    name = "s2"
+
+    failover = {
+      "*" = {
+        service        = "s3"
+        service_subset = "v1"
+      }
+    }
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-svc-failover/s3-v1.hcl

@@ -0,0 +1,13 @@
+services {
+  id   = "s3-v1"
+  name = "s3"
+  port = 8283
+
+  meta {
+    version = "v1"
+  }
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-svc-failover/s3-v2.hcl

@@ -0,0 +1,13 @@
+services {
+  id   = "s3-v2"
+  name = "s3"
+  port = 8284
+
+  meta {
+    version = "v2"
+  }
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-svc-failover/s3.hcl

@@ -0,0 +1,8 @@
+services {
+  name = "s3"
+  port = 8282
+
+  connect {
+    sidecar_service {}
+  }
+}

test/integration/connect/envoy/case-cfg-resolver-svc-failover/setup.sh

@@ -0,0 +1,23 @@
+#!/bin/bash
+
+set -euo pipefail
+
+# wait for bootstrap to apply config entries
+wait_for_config_entry proxy-defaults global
+wait_for_config_entry service-resolver s2
+wait_for_config_entry service-resolver s3
+
+# s2, s3, and s3-v1 are retained just to have a honeypot for bad envoy configs to route into
+gen_envoy_bootstrap s1 19000
+gen_envoy_bootstrap s2 19001
+gen_envoy_bootstrap s3-v1 19002
+gen_envoy_bootstrap s3-v2 19003
+gen_envoy_bootstrap s3 19004
+
+export REQUIRED_SERVICES="
+s1 s1-sidecar-proxy
+s2 s2-sidecar-proxy
+s3 s3-sidecar-proxy
+s3-v1 s3-v1-sidecar-proxy
+s3-v2 s3-v2-sidecar-proxy
+"

test/integration/connect/envoy/case-cfg-resolver-svc-failover/verify.bats

@@ -0,0 +1,75 @@
+#!/usr/bin/env bats
+
+load helpers
+
+@test "s1 proxy admin is up on :19000" {
+  retry_default curl -f -s localhost:19000/stats -o /dev/null
+}
+
+@test "s2 proxy admin is up on :19001" {
+  retry_default curl -f -s localhost:19001/stats -o /dev/null
+}
+
+@test "s3-v1 proxy admin is up on :19002" {
+  retry_default curl -f -s localhost:19002/stats -o /dev/null
+}
+
+@test "s3-v2 proxy admin is up on :19003" {
+  retry_default curl -f -s localhost:19003/stats -o /dev/null
+}
+
+@test "s3 proxy admin is up on :19004" {
+  retry_default curl -f -s localhost:19004/stats -o /dev/null
+}
+
+@test "s1 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21000 s1
+}
+
+@test "s2 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21001 s2
+}
+
+@test "s3-v1 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21002 s3
+}
+
+@test "s3-v2 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21003 s3
+}
+
+@test "s3 proxy listener should be up and have right cert" {
+  assert_proxy_presents_cert_uri localhost:21004 s3
+}
+
+@test "s2 proxies should be healthy" {
+  assert_service_has_healthy_instances s2 1
+}
+
+@test "s3 proxies should be healthy" {
+  assert_service_has_healthy_instances s3 3
+}
+
+# Note: when failover is configured the cluster is named for the original
+# service not any destination related to failover.
+@test "s1 upstream should have healthy endpoints for s2 and s3 together" {
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 2
+}
+
+@test "s1 upstream should be able to connect to s2 via upstream s2 to start" {
+  assert_expected_fortio_name s2
+}
+
+@test "terminate instance of s2 envoy which should trigger failover to s3 when tcp check fails" {
+  kill_envoy s2
+}
+
+@test "s1 upstream should have healthy endpoints for s3-v1 and unhealthy endpoints for s2" {
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 UNHEALTHY 1
+}
+
+@test "s1 upstream should be able to connect to s3-v1 now" {
+  assert_expected_fortio_name s3-v1
+}
+

test/integration/connect/envoy/case-cfg-resolver-svc-redirect/setup.sh

@@ -2,9 +2,16 @@
 
 set -euo pipefail
 
-# retry because resolving the central config might race
-retry_default gen_envoy_bootstrap s1 19000
-retry_default gen_envoy_bootstrap s2 19001
-retry_default gen_envoy_bootstrap s3 19002
+# wait for bootstrap to apply config entries
+wait_for_config_entry proxy-defaults global
+wait_for_config_entry service-resolver s2
 
-export REQUIRED_SERVICES="s1 s1-sidecar-proxy s2 s2-sidecar-proxy s3 s3-sidecar-proxy"
+gen_envoy_bootstrap s1 19000
+gen_envoy_bootstrap s2 19001
+gen_envoy_bootstrap s3 19002
+
+export REQUIRED_SERVICES="
+s1 s1-sidecar-proxy
+s2 s2-sidecar-proxy
+s3 s3-sidecar-proxy
+"

test/integration/connect/envoy/case-cfg-resolver-svc-redirect/verify.bats

@@ -31,7 +31,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s3" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s3 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s3 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to its upstream simply" {

test/integration/connect/envoy/case-dogstatsd-udp/verify.bats

@@ -15,7 +15,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2" {

test/integration/connect/envoy/case-grpc/verify.bats

@@ -15,7 +15,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2 via grpc" {

test/integration/connect/envoy/case-http-badauthz/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should NOT be able to connect to s2" {

test/integration/connect/envoy/case-http/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2 with http/1.1" {

test/integration/connect/envoy/case-http2/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2 via http2" {

test/integration/connect/envoy/case-prometheus/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2 with http/1.1" {

test/integration/connect/envoy/case-statsd-udp/verify.bats

@@ -15,7 +15,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2" {

test/integration/connect/envoy/case-zipkin/verify.bats

@@ -23,7 +23,7 @@ load helpers
 }
 
 @test "s1 upstream should have healthy endpoints for s2" {
-  assert_upstream_has_healthy_endpoints 127.0.0.1:19000 s2 1
+  assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2 HEALTHY 1
 }
 
 @test "s1 upstream should be able to connect to s2" {

test/integration/connect/envoy/docker-compose.yml

@@ -74,6 +74,38 @@ services:
      - "disabled"
     network_mode: service:consul
 
+  s2-v1:
+    depends_on:
+      - consul
+    image: "fortio/fortio"
+    environment:
+      - "FORTIO_NAME=s2-v1"
+    command:
+     - "server"
+     - "-http-port"
+     - ":8182"
+     - "-grpc-port"
+     - ":8178"
+     - "-redirect-port"
+     - "disabled"
+    network_mode: service:consul
+
+  s2-v2:
+    depends_on:
+      - consul
+    image: "fortio/fortio"
+    environment:
+      - "FORTIO_NAME=s2-v2"
+    command:
+     - "server"
+     - "-http-port"
+     - ":8183"
+     - "-grpc-port"
+     - ":8177"
+     - "-redirect-port"
+     - "disabled"
+    network_mode: service:consul
+
   s3:
     depends_on:
       - consul
@@ -90,6 +122,38 @@ services:
      - "disabled"
     network_mode: service:consul
 
+  s3-v1:
+    depends_on:
+      - consul
+    image: "fortio/fortio"
+    environment:
+      - "FORTIO_NAME=s3-v1"
+    command:
+     - "server"
+     - "-http-port"
+     - ":8283"
+     - "-grpc-port"
+     - ":8278"
+     - "-redirect-port"
+     - "disabled"
+    network_mode: service:consul
+
+  s3-v2:
+    depends_on:
+      - consul
+    image: "fortio/fortio"
+    environment:
+      - "FORTIO_NAME=s3-v2"
+    command:
+     - "server"
+     - "-http-port"
+     - ":8284"
+     - "-grpc-port"
+     - ":8277"
+     - "-redirect-port"
+     - "disabled"
+    network_mode: service:consul
+
   s1-sidecar-proxy:
     depends_on:
       - consul
@@ -132,6 +196,48 @@ services:
       - *workdir-volume
     network_mode: service:consul
 
+  s2-v1-sidecar-proxy:
+    depends_on:
+      - consul
+    image: "envoyproxy/envoy:v${ENVOY_VERSION:-1.8.0}"
+    command:
+     - "envoy"
+     - "-c"
+     - "/workdir/envoy/s2-v1-bootstrap.json"
+     - "-l"
+     - "debug"
+     # Hot restart breaks since both envoys seem to interact with each other
+     # despite separate containers that don't share IPC namespace. Not quite
+     # sure how this happens but may be due to unix socket being in some shared
+     # location?
+     - "--disable-hot-restart"
+     - "--drain-time-s"
+     - "1"
+    volumes:
+      - *workdir-volume
+    network_mode: service:consul
+
+  s2-v2-sidecar-proxy:
+    depends_on:
+      - consul
+    image: "envoyproxy/envoy:v${ENVOY_VERSION:-1.8.0}"
+    command:
+     - "envoy"
+     - "-c"
+     - "/workdir/envoy/s2-v2-bootstrap.json"
+     - "-l"
+     - "debug"
+     # Hot restart breaks since both envoys seem to interact with each other
+     # despite separate containers that don't share IPC namespace. Not quite
+     # sure how this happens but may be due to unix socket being in some shared
+     # location?
+     - "--disable-hot-restart"
+     - "--drain-time-s"
+     - "1"
+    volumes:
+      - *workdir-volume
+    network_mode: service:consul
+
   s3-sidecar-proxy:
     depends_on:
       - consul
@@ -153,6 +259,48 @@ services:
       - *workdir-volume
     network_mode: service:consul
 
+  s3-v1-sidecar-proxy:
+    depends_on:
+      - consul
+    image: "envoyproxy/envoy:v${ENVOY_VERSION:-1.8.0}"
+    command:
+     - "envoy"
+     - "-c"
+     - "/workdir/envoy/s3-v1-bootstrap.json"
+     - "-l"
+     - "debug"
+     # Hot restart breaks since both envoys seem to interact with each other
+     # despite separate containers that don't share IPC namespace. Not quite
+     # sure how this happens but may be due to unix socket being in some shared
+     # location?
+     - "--disable-hot-restart"
+     - "--drain-time-s"
+     - "1"
+    volumes:
+      - *workdir-volume
+    network_mode: service:consul
+
+  s3-v2-sidecar-proxy:
+    depends_on:
+      - consul
+    image: "envoyproxy/envoy:v${ENVOY_VERSION:-1.8.0}"
+    command:
+     - "envoy"
+     - "-c"
+     - "/workdir/envoy/s3-v2-bootstrap.json"
+     - "-l"
+     - "debug"
+     # Hot restart breaks since both envoys seem to interact with each other
+     # despite separate containers that don't share IPC namespace. Not quite
+     # sure how this happens but may be due to unix socket being in some shared
+     # location?
+     - "--disable-hot-restart"
+     - "--drain-time-s"
+     - "1"
+    volumes:
+      - *workdir-volume
+    network_mode: service:consul
+
   verify:
     depends_on:
       - consul
@@ -168,6 +316,7 @@ services:
     volumes:
       - *workdir-volume
     network_mode: service:consul
+    pid: host
 
   s1-sidecar-proxy-consul-exec:
     depends_on:

test/integration/connect/envoy/helpers.bash

@@ -28,6 +28,10 @@ function retry_default {
   retry 5 1 $@
 }
 
+function retry_long {
+  retry 30 1 $@
+}
+
 function echored {
   tput setaf 1
   tput bold
@@ -120,34 +124,38 @@ function snapshot_envoy_admin {
   docker_wget "http://${HOSTPORT}/clusters?format=json" -q -O - > "./workdir/envoy/${ENVOY_NAME}-clusters.json"
 }
 
-function get_healthy_upstream_endpoint_count {
+function get_upstream_endpoint_in_status_count {
   local HOSTPORT=$1
   local CLUSTER_NAME=$2
+  local HEALTH_STATUS=$3
   run retry_default curl -s -f "http://${HOSTPORT}/clusters?format=json"
   [ "$status" -eq 0 ]
+  # echo "$output" >&3
   echo "$output" | jq --raw-output "
 .cluster_statuses[]
 | select(.name|startswith(\"${CLUSTER_NAME}.default.dc1.internal.\"))
-| .host_statuses[].health_status
-| select(.eds_health_status == \"HEALTHY\")
+| [.host_statuses[].health_status.eds_health_status]
+| [select(.[] == \"${HEALTH_STATUS}\")]
 | length"
 }
 
-function assert_upstream_has_healthy_endpoints_once {
+function assert_upstream_has_endpoints_in_status_once {
   local HOSTPORT=$1
   local CLUSTER_NAME=$2
-  local EXPECT_COUNT=$3
+  local HEALTH_STATUS=$3
+  local EXPECT_COUNT=$4
 
-  GOT_COUNT=$(get_healthy_upstream_endpoint_count $HOSTPORT $CLUSTER_NAME)
+  GOT_COUNT=$(get_upstream_endpoint_in_status_count $HOSTPORT $CLUSTER_NAME $HEALTH_STATUS)
 
   [ "$GOT_COUNT" -eq $EXPECT_COUNT ]
 }
 
-function assert_upstream_has_healthy_endpoints {
+function assert_upstream_has_endpoints_in_status {
   local HOSTPORT=$1
   local CLUSTER_NAME=$2
-  local EXPECT_COUNT=$3
-  run retry 30 1 assert_upstream_has_healthy_endpoints_once $HOSTPORT $CLUSTER_NAME $EXPECT_COUNT
+  local HEALTH_STATUS=$3
+  local EXPECT_COUNT=$4
+  run retry_long assert_upstream_has_endpoints_in_status_once $HOSTPORT $CLUSTER_NAME $HEALTH_STATUS $EXPECT_COUNT
   [ "$status" -eq 0 ]
 }
 
@@ -171,18 +179,37 @@ function assert_service_has_healthy_instances {
   local SERVICE_NAME=$1
   local EXPECT_COUNT=$2
 
-  run retry 30 1 assert_service_has_healthy_instances_once $SERVICE_NAME $EXPECT_COUNT
+  run retry_long assert_service_has_healthy_instances_once $SERVICE_NAME $EXPECT_COUNT
   [ "$status" -eq 0 ]
 }
 
 function docker_consul {
-  docker run -ti --rm --network container:envoy_consul_1 consul-dev $@
+  docker run -i --rm --network container:envoy_consul_1 consul-dev $@
 }
 
 function docker_wget {
   docker run -ti --rm --network container:envoy_consul_1 alpine:3.9 wget $@
 }
 
+function get_envoy_pid {
+  local BOOTSTRAP_NAME=$1
+  run ps aux
+  [ "$status" == 0 ]
+  PID="$(echo "$output" | grep "envoy -c /workdir/envoy/${BOOTSTRAP_NAME}-bootstrap.json" | awk '{print $1}')"
+  [ -n "$PID" ]
+
+  echo "$PID"
+}
+
+function kill_envoy {
+  local BOOTSTRAP_NAME=$1
+
+  PID="$(get_envoy_pid $BOOTSTRAP_NAME)"
+  echo "PID = $PID"
+
+  kill -TERM $PID
+}
+
 function must_match_in_statsd_logs {
   run cat /workdir/statsd/statsd.log
   COUNT=$( echo "$output" | grep -Ec $1 )
@@ -254,6 +281,18 @@ function gen_envoy_bootstrap {
   fi
 }
 
+function read_config_entry {
+  local KIND=$1
+  local NAME=$2
+  docker_consul config read -kind $KIND -name $NAME
+}
+
+function wait_for_config_entry {
+  local KIND=$1
+  local NAME=$2
+  retry_default read_config_entry $KIND $NAME >/dev/null
+}
+
 function get_upstream_fortio_name {
   run retry_default curl -v -s -f localhost:5000/debug?env=dump
   [ "$status" == 0 ]