open-consul/test/integration/connect/envoy/run-tests.sh

904 lines
25 KiB
Bash
Raw Normal View History

#!/usr/bin/env bash
set -eEuo pipefail
readonly self_name="$0"
readonly HASHICORP_DOCKER_PROXY="docker.mirror.hashicorp.services"
# DEBUG=1 enables set -x for this script so echos every command run
DEBUG=${DEBUG:-}
XDS_TARGET=${XDS_TARGET:-server}
# ENVOY_VERSION to run each test against
ENVOY_VERSION=${ENVOY_VERSION:-"1.23.1"}
export ENVOY_VERSION
export DOCKER_BUILDKIT=1
# Always run tests on amd64 because that's what the CI environment uses.
export DOCKER_DEFAULT_PLATFORM="linux/amd64"
if [ ! -z "$DEBUG" ] ; then
set -x
fi
source helpers.bash
function command_error {
echo "ERR: command exited with status $1" 1>&2
echo " command: $2" 1>&2
echo " line: $3" 1>&2
echo " function: $4" 1>&2
echo " called at: $5" 1>&2
# printf '%s\n' "${FUNCNAME[@]}"
# printf '%s\n' "${BASH_SOURCE[@]}"
# printf '%s\n' "${BASH_LINENO[@]}"
}
trap 'command_error $? "${BASH_COMMAND}" "${LINENO}" "${FUNCNAME[0]:-main}" "${BASH_SOURCE[0]}:${BASH_LINENO[0]}"' ERR
readonly WORKDIR_SNIPPET='-v envoy_workdir:/workdir'
function network_snippet {
local DC="$1"
echo "--net container:envoy_consul-${DC}_1"
}
function aws_snippet {
LAMBDA_TESTS_ENABLED=${LAMBDA_TESTS_ENABLED:-false}
if [ "$LAMBDA_TESTS_ENABLED" != false ]; then
local snippet=""
# The Lambda integration cases assume that a Lambda function exists in $AWS_REGION with an ARN of $AWS_LAMBDA_ARN.
# The AWS credentials must have permission to invoke the Lambda function.
[ -n "$(set | grep '^AWS_ACCESS_KEY_ID=')" ] && snippet="${snippet} -e AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID"
[ -n "$(set | grep '^AWS_SECRET_ACCESS_KEY=')" ] && snippet="${snippet} -e AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY"
[ -n "$(set | grep '^AWS_SESSION_TOKEN=')" ] && snippet="${snippet} -e AWS_SESSION_TOKEN=$AWS_SESSION_TOKEN"
[ -n "$(set | grep '^AWS_LAMBDA_REGION=')" ] && snippet="${snippet} -e AWS_LAMBDA_REGION=$AWS_LAMBDA_REGION"
[ -n "$(set | grep '^AWS_LAMBDA_ARN=')" ] && snippet="${snippet} -e AWS_LAMBDA_ARN=$AWS_LAMBDA_ARN"
echo "$snippet"
fi
}
function init_workdir {
2021-11-12 21:45:50 +00:00
local CLUSTER="$1"
2021-11-12 21:45:50 +00:00
if test -z "$CLUSTER"
then
2021-11-12 21:45:50 +00:00
CLUSTER=primary
fi
# Note, we use explicit set of dirs so we don't delete .gitignore. Also,
# don't wipe logs between runs as they are already split and we need them to
# upload as artifacts later.
2021-11-12 21:45:50 +00:00
rm -rf workdir/${CLUSTER}
mkdir -p workdir/${CLUSTER}/{consul,consul-server,register,envoy,bats,statsd,data}
# Reload consul config from defaults
2021-11-12 21:45:50 +00:00
cp consul-base-cfg/*.hcl workdir/${CLUSTER}/consul/
# Add any overrides if there are any (no op if not)
2021-11-12 21:45:50 +00:00
find ${CASE_DIR} -maxdepth 1 -name '*.hcl' -type f -exec cp -f {} workdir/${CLUSTER}/consul \;
# Copy all the test files
2021-11-12 21:45:50 +00:00
find ${CASE_DIR} -maxdepth 1 -name '*.bats' -type f -exec cp -f {} workdir/${CLUSTER}/bats \;
# Copy CLUSTER specific bats
cp helpers.bash workdir/${CLUSTER}/bats
2021-11-12 21:45:50 +00:00
# Add any CLUSTER overrides
if test -d "${CASE_DIR}/${CLUSTER}"
then
2021-11-12 21:45:50 +00:00
find ${CASE_DIR}/${CLUSTER} -type f -name '*.hcl' -exec cp -f {} workdir/${CLUSTER}/consul \;
find ${CASE_DIR}/${CLUSTER} -type f -name '*.bats' -exec cp -f {} workdir/${CLUSTER}/bats \;
fi
# move all of the registration files OUT of the consul config dir now
2021-11-12 21:45:50 +00:00
find workdir/${CLUSTER}/consul -type f -name 'service_*.hcl' -exec mv -f {} workdir/${CLUSTER}/register \;
# move the server.hcl out of the consul dir so that it doesn't get picked up
# by the client agent (if we're running with XDS_TARGET=client).
if test -f "workdir/${CLUSTER}/consul/server.hcl"
then
mv workdir/${CLUSTER}/consul/server.hcl workdir/${CLUSTER}/consul-server/server.hcl
fi
if test -f "workdir/${CLUSTER}/consul/peering_server.hcl" -a $REQUIRE_PEERS = "1"
then
mv workdir/${CLUSTER}/consul/peering_server.hcl workdir/${CLUSTER}/consul-server/peering_server.hcl
else
rm workdir/${CLUSTER}/consul/peering_server.hcl
fi
# copy the ca-certs for SDS so we can verify the right ones are served
mkdir -p workdir/test-sds-server/certs
cp test-sds-server/certs/ca-root.crt workdir/test-sds-server/certs/ca-root.crt
if test -d "${CASE_DIR}/data"
then
2021-11-12 21:45:50 +00:00
cp -r ${CASE_DIR}/data/* workdir/${CLUSTER}/data
fi
return 0
}
function docker_kill_rm {
local name
local todo=()
for name in "$@"; do
name="envoy_${name}_1"
if docker container inspect $name &>/dev/null; then
if [[ "$name" == envoy_tcpdump-* ]]; then
echo -n "Gracefully stopping $name..."
docker stop $name &> /dev/null
echo "done"
fi
todo+=($name)
fi
done
if [[ ${#todo[@]} -eq 0 ]]; then
return 0
fi
echo -n "Killing and removing: ${todo[@]}..."
docker rm -v -f ${todo[@]} &> /dev/null
echo "done"
}
function start_consul {
local DC=${1:-primary}
# 8500/8502 are for consul
# 9411 is for zipkin which shares the network with consul
# 16686 is for jaeger ui which also shares the network with consul
ports=(
'-p=8500:8500'
'-p=8502:8502'
'-p=9411:9411'
'-p=16686:16686'
)
case "$DC" in
secondary)
ports=(
'-p=9500:8500'
'-p=9502:8502'
)
;;
alpha)
ports=(
'-p=9510:8500'
'-p=9512:8502'
)
;;
esac
license="${CONSUL_LICENSE:-}"
# load the consul license so we can pass it into the consul
# containers as an env var in the case that this is a consul
# enterprise test
if test -z "$license" -a -n "${CONSUL_LICENSE_PATH:-}"
then
license=$(cat $CONSUL_LICENSE_PATH)
fi
# We currently run these integration tests in two modes: one in which Envoy's
# xDS sessions are served directly by a Consul server, and another in which it
# goes through a client agent.
#
# This is necessary because servers and clients source configuration data in
# different ways (client agents use an RPC-backed cache and servers use their
# own local data) and we want to catch regressions in both.
#
# In the future we should also expand these tests to register services to the
# catalog directly (agentless) rather than relying on the server also being
# an agent.
#
# When XDS_TARGET=client we'll start a Consul server with its gRPC port
# disabled (but only if REQUIRE_PEERS is not set), and a client agent with
# its gRPC port enabled.
#
# When XDS_TARGET=server (or anything else) we'll run a single Consul server
# with its gRPC port enabled.
#
# In either case, the hostname `consul-${DC}-server` should be used as a
# server address (e.g. for WAN joining) and `consul-${DC}-client` should be
# used as a client address (e.g. for interacting with the HTTP API).
#
# Both hostnames work in both modes because we set network aliases on the
# containers such that both hostnames will resolve to the same container when
# XDS_TARGET=server.
#
# We also join containers to the network `container:consul-${DC}_1` in many
# places (see: network_snippet) so that we can curl localhost etc. In both
# modes, you can assume that this name refers to the client's container.
#
# Any .hcl files in the case/cluster directory will be given to both clients
# and servers (via the -config-dir flag) *except for* server.hcl which will
# only be applied to the server (and service registrations which will be made
# against the client).
if [[ "$XDS_TARGET" == "client" ]]
then
docker_kill_rm consul-${DC}-server
docker_kill_rm consul-${DC}
docker run -d --name envoy_consul-${DC}-server_1 \
--net=envoy-tests \
$WORKDIR_SNIPPET \
--hostname "consul-${DC}-server" \
--network-alias "consul-${DC}-server" \
-e "CONSUL_LICENSE=$license" \
consul:local \
agent -dev -datacenter "${DC}" \
-config-dir "/workdir/${DC}/consul" \
-config-dir "/workdir/${DC}/consul-server" \
-client "0.0.0.0" \
-bind "0.0.0.0" >/dev/null
docker run -d --name envoy_consul-${DC}_1 \
--net=envoy-tests \
$WORKDIR_SNIPPET \
--hostname "consul-${DC}-client" \
--network-alias "consul-${DC}-client" \
-e "CONSUL_LICENSE=$license" \
${ports[@]} \
consul:local \
agent -datacenter "${DC}" \
-config-dir "/workdir/${DC}/consul" \
-data-dir "/tmp/consul" \
-client "0.0.0.0" \
-grpc-port 8502 \
-datacenter "${DC}" \
-retry-join "consul-${DC}-server" >/dev/null
else
docker_kill_rm consul-${DC}
docker run -d --name envoy_consul-${DC}_1 \
--net=envoy-tests \
$WORKDIR_SNIPPET \
--hostname "consul-${DC}" \
--network-alias "consul-${DC}-client" \
--network-alias "consul-${DC}-server" \
-e "CONSUL_LICENSE=$license" \
${ports[@]} \
consul:local \
agent -dev -datacenter "${DC}" \
-config-dir "/workdir/${DC}/consul" \
-config-dir "/workdir/${DC}/consul-server" \
-client "0.0.0.0" >/dev/null
fi
}
2021-11-12 21:45:50 +00:00
function start_partitioned_client {
local PARTITION=${1:-ap1}
# Start consul now as setup script needs it up
docker_kill_rm consul-${PARTITION}
license="${CONSUL_LICENSE:-}"
# load the consul license so we can pass it into the consul
# containers as an env var in the case that this is a consul
# enterprise test
if test -z "$license" -a -n "${CONSUL_LICENSE_PATH:-}"
then
license=$(cat $CONSUL_LICENSE_PATH)
fi
sh -c "rm -rf /workdir/${PARTITION}/data"
# Run consul and expose some ports to the host to make debugging locally a
# bit easier.
#
docker run -d --name envoy_consul-${PARTITION}_1 \
--net=envoy-tests \
$WORKDIR_SNIPPET \
--hostname "consul-${PARTITION}-client" \
--network-alias "consul-${PARTITION}-client" \
2021-11-12 21:45:50 +00:00
-e "CONSUL_LICENSE=$license" \
consul:local agent \
2021-11-12 21:45:50 +00:00
-datacenter "primary" \
-retry-join "consul-primary-server" \
2021-11-12 21:45:50 +00:00
-grpc-port 8502 \
-data-dir "/tmp/consul" \
-config-dir "/workdir/${PARTITION}/consul" \
-client "0.0.0.0" >/dev/null
}
function pre_service_setup {
2021-11-12 21:45:50 +00:00
local CLUSTER=${1:-primary}
# Run test case setup (e.g. generating Envoy bootstrap, starting containers)
2021-11-12 21:45:50 +00:00
if [ -f "${CASE_DIR}/${CLUSTER}/setup.sh" ]
then
2021-11-12 21:45:50 +00:00
source ${CASE_DIR}/${CLUSTER}/setup.sh
else
source ${CASE_DIR}/setup.sh
fi
}
function start_services {
# Push the state to the shared docker volume (note this is because CircleCI
# can't use shared volumes)
docker cp workdir/. envoy_workdir_1:/workdir
# Start containers required
if [ ! -z "$REQUIRED_SERVICES" ] ; then
docker_kill_rm $REQUIRED_SERVICES
run_containers $REQUIRED_SERVICES
fi
return 0
}
function verify {
2021-11-12 21:45:50 +00:00
local CLUSTER="$1"
if test -z "$CLUSTER"; then
CLUSTER="primary"
fi
# Execute tests
res=0
# Nuke any previous case's verify container.
2021-11-12 21:45:50 +00:00
docker_kill_rm verify-${CLUSTER}
2021-11-12 21:45:50 +00:00
echo "Running ${CLUSTER} verification step for ${CASE_DIR}..."
# need to tell the PID 1 inside of the container that it won't be actual PID
# 1 because we're using --pid=host so we use TINI_SUBREAPER
2021-11-12 21:45:50 +00:00
if docker run --name envoy_verify-${CLUSTER}_1 -t \
-e TINI_SUBREAPER=1 \
-e ENVOY_VERSION \
$WORKDIR_SNIPPET \
--pid=host \
2021-11-12 21:45:50 +00:00
$(network_snippet $CLUSTER) \
$(aws_snippet) \
bats-verify \
Remove terminal colouring from test output so it is (#14810) more readable in CI. ``` Running primary verification step for case-ingress-gateway-multiple-services... �[34;1mverify.bats �[0m�[1G ingress proxy admin is up on :20000�[K�[75G 1/12�[2G�[1G ✓ ingress proxy admin is up on :20000�[K �[0m�[1G s1 proxy admin is up on :19000�[K�[75G 2/12�[2G�[1G ✓ s1 proxy admin is up on :19000�[K �[0m�[1G s2 proxy admin is up on :19001�[K�[75G 3/12�[2G�[1G ✓ s2 proxy admin is up on :19001�[K �[0m�[1G s1 proxy listener should be up and have right cert�[K�[75G 4/12�[2G�[1G ✓ s1 proxy listener should be up and have right cert�[K �[0m�[1G s2 proxy listener should be up and have right cert�[K�[75G 5/12�[2G�[1G ✓ s2 proxy listener should be up and have right cert�[K �[0m�[1G ingress-gateway should have healthy endpoints for s1�[K�[75G 6/12�[2G�[31;1m�[1G ✗ ingress-gateway should have healthy endpoints for s1�[K �[0m�[31;22m (from function `assert_upstream_has_endpoints_in_status' in file /workdir/primary/bats/helpers.bash, line 385, ``` versus ``` Running primary verification step for case-ingress-gateway-multiple-services... 1..12 ok 1 ingress proxy admin is up on :20000 ok 2 s1 proxy admin is up on :19000 ok 3 s2 proxy admin is up on :19001 ok 4 s1 proxy listener should be up and have right cert ok 5 s2 proxy listener should be up and have right cert not ok 6 ingress-gateway should have healthy endpoints for s1 not ok 7 s1 proxy should have been configured with max_connections in services ok 8 ingress-gateway should have healthy endpoints for s2 ```
2022-10-04 15:35:19 +00:00
--formatter tap /workdir/${CLUSTER}/bats ; then
echo "✓ PASS"
else
Remove terminal colouring from test output so it is (#14810) more readable in CI. ``` Running primary verification step for case-ingress-gateway-multiple-services... �[34;1mverify.bats �[0m�[1G ingress proxy admin is up on :20000�[K�[75G 1/12�[2G�[1G ✓ ingress proxy admin is up on :20000�[K �[0m�[1G s1 proxy admin is up on :19000�[K�[75G 2/12�[2G�[1G ✓ s1 proxy admin is up on :19000�[K �[0m�[1G s2 proxy admin is up on :19001�[K�[75G 3/12�[2G�[1G ✓ s2 proxy admin is up on :19001�[K �[0m�[1G s1 proxy listener should be up and have right cert�[K�[75G 4/12�[2G�[1G ✓ s1 proxy listener should be up and have right cert�[K �[0m�[1G s2 proxy listener should be up and have right cert�[K�[75G 5/12�[2G�[1G ✓ s2 proxy listener should be up and have right cert�[K �[0m�[1G ingress-gateway should have healthy endpoints for s1�[K�[75G 6/12�[2G�[31;1m�[1G ✗ ingress-gateway should have healthy endpoints for s1�[K �[0m�[31;22m (from function `assert_upstream_has_endpoints_in_status' in file /workdir/primary/bats/helpers.bash, line 385, ``` versus ``` Running primary verification step for case-ingress-gateway-multiple-services... 1..12 ok 1 ingress proxy admin is up on :20000 ok 2 s1 proxy admin is up on :19000 ok 3 s2 proxy admin is up on :19001 ok 4 s1 proxy listener should be up and have right cert ok 5 s2 proxy listener should be up and have right cert not ok 6 ingress-gateway should have healthy endpoints for s1 not ok 7 s1 proxy should have been configured with max_connections in services ok 8 ingress-gateway should have healthy endpoints for s2 ```
2022-10-04 15:35:19 +00:00
echo " FAIL"
res=1
fi
return $res
}
function capture_logs {
local LOG_DIR="workdir/logs/${CASE_DIR}/${ENVOY_VERSION}"
init_vars
echo "Capturing Logs"
mkdir -p "$LOG_DIR"
services="$REQUIRED_SERVICES consul-primary"
if [[ "$XDS_TARGET" == "client" ]]
then
services="$services consul-primary-server"
fi
if is_set $REQUIRE_SECONDARY
then
services="$services consul-secondary"
if [[ "$XDS_TARGET" == "client" ]]
then
services="$services consul-secondary-server"
fi
fi
2021-11-12 21:45:50 +00:00
if is_set $REQUIRE_PARTITIONS
then
services="$services consul-ap1"
fi
if is_set $REQUIRE_PEERS
then
services="$services consul-alpha"
if [[ "$XDS_TARGET" == "client" ]]
then
services="$services consul-alpha-server"
fi
fi
if [ -f "${CASE_DIR}/capture.sh" ]
then
echo "Executing ${CASE_DIR}/capture.sh"
source ${CASE_DIR}/capture.sh || true
fi
for cont in $services; do
echo "Capturing log for $cont"
docker logs "envoy_${cont}_1" &> "${LOG_DIR}/${cont}.log" || {
echo "EXIT CODE $?" > "${LOG_DIR}/${cont}.log"
}
done
}
function stop_services {
# Teardown
docker_kill_rm $REQUIRED_SERVICES
docker_kill_rm consul-primary consul-primary-server consul-secondary consul-secondary-server consul-ap1 consul-alpha consul-alpha-server
}
function init_vars {
source "defaults.sh"
if [ -f "${CASE_DIR}/vars.sh" ] ; then
source "${CASE_DIR}/vars.sh"
fi
}
wan federation via mesh gateways (#6884) This is like a Möbius strip of code due to the fact that low-level components (serf/memberlist) are connected to high-level components (the catalog and mesh-gateways) in a twisty maze of references which make it hard to dive into. With that in mind here's a high level summary of what you'll find in the patch: There are several distinct chunks of code that are affected: * new flags and config options for the server * retry join WAN is slightly different * retry join code is shared to discover primary mesh gateways from secondary datacenters * because retry join logic runs in the *agent* and the results of that operation for primary mesh gateways are needed in the *server* there are some methods like `RefreshPrimaryGatewayFallbackAddresses` that must occur at multiple layers of abstraction just to pass the data down to the right layer. * new cache type `FederationStateListMeshGatewaysName` for use in `proxycfg/xds` layers * the function signature for RPC dialing picked up a new required field (the node name of the destination) * several new RPCs for manipulating a FederationState object: `FederationState:{Apply,Get,List,ListMeshGateways}` * 3 read-only internal APIs for debugging use to invoke those RPCs from curl * raft and fsm changes to persist these FederationStates * replication for FederationStates as they are canonically stored in the Primary and replicated to the Secondaries. * a special derivative of anti-entropy that runs in secondaries to snapshot their local mesh gateway `CheckServiceNodes` and sync them into their upstream FederationState in the primary (this works in conjunction with the replication to distribute addresses for all mesh gateways in all DCs to all other DCs) * a "gateway locator" convenience object to make use of this data to choose the addresses of gateways to use for any given RPC or gossip operation to a remote DC. This gets data from the "retry join" logic in the agent and also directly calls into the FSM. * RPC (`:8300`) on the server sniffs the first byte of a new connection to determine if it's actually doing native TLS. If so it checks the ALPN header for protocol determination (just like how the existing system uses the type-byte marker). * 2 new kinds of protocols are exclusively decoded via this native TLS mechanism: one for ferrying "packet" operations (udp-like) from the gossip layer and one for "stream" operations (tcp-like). The packet operations re-use sockets (using length-prefixing) to cut down on TLS re-negotiation overhead. * the server instances specially wrap the `memberlist.NetTransport` when running with gateway federation enabled (in a `wanfed.Transport`). The general gist is that if it tries to dial a node in the SAME datacenter (deduced by looking at the suffix of the node name) there is no change. If dialing a DIFFERENT datacenter it is wrapped up in a TLS+ALPN blob and sent through some mesh gateways to eventually end up in a server's :8300 port. * a new flag when launching a mesh gateway via `consul connect envoy` to indicate that the servers are to be exposed. This sets a special service meta when registering the gateway into the catalog. * `proxycfg/xds` notice this metadata blob to activate additional watches for the FederationState objects as well as the location of all of the consul servers in that datacenter. * `xds:` if the extra metadata is in place additional clusters are defined in a DC to bulk sink all traffic to another DC's gateways. For the current datacenter we listen on a wildcard name (`server.<dc>.consul`) that load balances all servers as well as one mini-cluster per node (`<node>.server.<dc>.consul`) * the `consul tls cert create` command got a new flag (`-node`) to help create an additional SAN in certs that can be used with this flavor of federation.
2020-03-09 20:59:02 +00:00
function global_setup {
if [ -f "${CASE_DIR}/global-setup.sh" ] ; then
source "${CASE_DIR}/global-setup.sh"
wan federation via mesh gateways (#6884) This is like a Möbius strip of code due to the fact that low-level components (serf/memberlist) are connected to high-level components (the catalog and mesh-gateways) in a twisty maze of references which make it hard to dive into. With that in mind here's a high level summary of what you'll find in the patch: There are several distinct chunks of code that are affected: * new flags and config options for the server * retry join WAN is slightly different * retry join code is shared to discover primary mesh gateways from secondary datacenters * because retry join logic runs in the *agent* and the results of that operation for primary mesh gateways are needed in the *server* there are some methods like `RefreshPrimaryGatewayFallbackAddresses` that must occur at multiple layers of abstraction just to pass the data down to the right layer. * new cache type `FederationStateListMeshGatewaysName` for use in `proxycfg/xds` layers * the function signature for RPC dialing picked up a new required field (the node name of the destination) * several new RPCs for manipulating a FederationState object: `FederationState:{Apply,Get,List,ListMeshGateways}` * 3 read-only internal APIs for debugging use to invoke those RPCs from curl * raft and fsm changes to persist these FederationStates * replication for FederationStates as they are canonically stored in the Primary and replicated to the Secondaries. * a special derivative of anti-entropy that runs in secondaries to snapshot their local mesh gateway `CheckServiceNodes` and sync them into their upstream FederationState in the primary (this works in conjunction with the replication to distribute addresses for all mesh gateways in all DCs to all other DCs) * a "gateway locator" convenience object to make use of this data to choose the addresses of gateways to use for any given RPC or gossip operation to a remote DC. This gets data from the "retry join" logic in the agent and also directly calls into the FSM. * RPC (`:8300`) on the server sniffs the first byte of a new connection to determine if it's actually doing native TLS. If so it checks the ALPN header for protocol determination (just like how the existing system uses the type-byte marker). * 2 new kinds of protocols are exclusively decoded via this native TLS mechanism: one for ferrying "packet" operations (udp-like) from the gossip layer and one for "stream" operations (tcp-like). The packet operations re-use sockets (using length-prefixing) to cut down on TLS re-negotiation overhead. * the server instances specially wrap the `memberlist.NetTransport` when running with gateway federation enabled (in a `wanfed.Transport`). The general gist is that if it tries to dial a node in the SAME datacenter (deduced by looking at the suffix of the node name) there is no change. If dialing a DIFFERENT datacenter it is wrapped up in a TLS+ALPN blob and sent through some mesh gateways to eventually end up in a server's :8300 port. * a new flag when launching a mesh gateway via `consul connect envoy` to indicate that the servers are to be exposed. This sets a special service meta when registering the gateway into the catalog. * `proxycfg/xds` notice this metadata blob to activate additional watches for the FederationState objects as well as the location of all of the consul servers in that datacenter. * `xds:` if the extra metadata is in place additional clusters are defined in a DC to bulk sink all traffic to another DC's gateways. For the current datacenter we listen on a wildcard name (`server.<dc>.consul`) that load balances all servers as well as one mini-cluster per node (`<node>.server.<dc>.consul`) * the `consul tls cert create` command got a new flag (`-node`) to help create an additional SAN in certs that can be used with this flavor of federation.
2020-03-09 20:59:02 +00:00
fi
}
function wipe_volumes {
docker run --rm -i \
$WORKDIR_SNIPPET \
--net=none \
"${HASHICORP_DOCKER_PROXY}/alpine" \
sh -c 'rm -rf /workdir/*'
}
function run_tests {
CASE_DIR="${CASE_DIR?CASE_DIR must be set to the path of the test case}"
CASE_NAME=$( basename $CASE_DIR | cut -c6- )
export CASE_NAME
export SKIP_CASE=""
init_vars
# Initialize the workdir
init_workdir primary
if is_set $REQUIRE_SECONDARY
then
init_workdir secondary
fi
2021-11-12 21:45:50 +00:00
if is_set $REQUIRE_PARTITIONS
then
init_workdir ap1
fi
if is_set $REQUIRE_PEERS
then
init_workdir alpha
fi
wan federation via mesh gateways (#6884) This is like a Möbius strip of code due to the fact that low-level components (serf/memberlist) are connected to high-level components (the catalog and mesh-gateways) in a twisty maze of references which make it hard to dive into. With that in mind here's a high level summary of what you'll find in the patch: There are several distinct chunks of code that are affected: * new flags and config options for the server * retry join WAN is slightly different * retry join code is shared to discover primary mesh gateways from secondary datacenters * because retry join logic runs in the *agent* and the results of that operation for primary mesh gateways are needed in the *server* there are some methods like `RefreshPrimaryGatewayFallbackAddresses` that must occur at multiple layers of abstraction just to pass the data down to the right layer. * new cache type `FederationStateListMeshGatewaysName` for use in `proxycfg/xds` layers * the function signature for RPC dialing picked up a new required field (the node name of the destination) * several new RPCs for manipulating a FederationState object: `FederationState:{Apply,Get,List,ListMeshGateways}` * 3 read-only internal APIs for debugging use to invoke those RPCs from curl * raft and fsm changes to persist these FederationStates * replication for FederationStates as they are canonically stored in the Primary and replicated to the Secondaries. * a special derivative of anti-entropy that runs in secondaries to snapshot their local mesh gateway `CheckServiceNodes` and sync them into their upstream FederationState in the primary (this works in conjunction with the replication to distribute addresses for all mesh gateways in all DCs to all other DCs) * a "gateway locator" convenience object to make use of this data to choose the addresses of gateways to use for any given RPC or gossip operation to a remote DC. This gets data from the "retry join" logic in the agent and also directly calls into the FSM. * RPC (`:8300`) on the server sniffs the first byte of a new connection to determine if it's actually doing native TLS. If so it checks the ALPN header for protocol determination (just like how the existing system uses the type-byte marker). * 2 new kinds of protocols are exclusively decoded via this native TLS mechanism: one for ferrying "packet" operations (udp-like) from the gossip layer and one for "stream" operations (tcp-like). The packet operations re-use sockets (using length-prefixing) to cut down on TLS re-negotiation overhead. * the server instances specially wrap the `memberlist.NetTransport` when running with gateway federation enabled (in a `wanfed.Transport`). The general gist is that if it tries to dial a node in the SAME datacenter (deduced by looking at the suffix of the node name) there is no change. If dialing a DIFFERENT datacenter it is wrapped up in a TLS+ALPN blob and sent through some mesh gateways to eventually end up in a server's :8300 port. * a new flag when launching a mesh gateway via `consul connect envoy` to indicate that the servers are to be exposed. This sets a special service meta when registering the gateway into the catalog. * `proxycfg/xds` notice this metadata blob to activate additional watches for the FederationState objects as well as the location of all of the consul servers in that datacenter. * `xds:` if the extra metadata is in place additional clusters are defined in a DC to bulk sink all traffic to another DC's gateways. For the current datacenter we listen on a wildcard name (`server.<dc>.consul`) that load balances all servers as well as one mini-cluster per node (`<node>.server.<dc>.consul`) * the `consul tls cert create` command got a new flag (`-node`) to help create an additional SAN in certs that can be used with this flavor of federation.
2020-03-09 20:59:02 +00:00
global_setup
# Allow vars.sh to set a reason to skip this test case based on the ENV
if [ "$SKIP_CASE" != "" ] ; then
Remove terminal colouring from test output so it is (#14810) more readable in CI. ``` Running primary verification step for case-ingress-gateway-multiple-services... �[34;1mverify.bats �[0m�[1G ingress proxy admin is up on :20000�[K�[75G 1/12�[2G�[1G ✓ ingress proxy admin is up on :20000�[K �[0m�[1G s1 proxy admin is up on :19000�[K�[75G 2/12�[2G�[1G ✓ s1 proxy admin is up on :19000�[K �[0m�[1G s2 proxy admin is up on :19001�[K�[75G 3/12�[2G�[1G ✓ s2 proxy admin is up on :19001�[K �[0m�[1G s1 proxy listener should be up and have right cert�[K�[75G 4/12�[2G�[1G ✓ s1 proxy listener should be up and have right cert�[K �[0m�[1G s2 proxy listener should be up and have right cert�[K�[75G 5/12�[2G�[1G ✓ s2 proxy listener should be up and have right cert�[K �[0m�[1G ingress-gateway should have healthy endpoints for s1�[K�[75G 6/12�[2G�[31;1m�[1G ✗ ingress-gateway should have healthy endpoints for s1�[K �[0m�[31;22m (from function `assert_upstream_has_endpoints_in_status' in file /workdir/primary/bats/helpers.bash, line 385, ``` versus ``` Running primary verification step for case-ingress-gateway-multiple-services... 1..12 ok 1 ingress proxy admin is up on :20000 ok 2 s1 proxy admin is up on :19000 ok 3 s2 proxy admin is up on :19001 ok 4 s1 proxy listener should be up and have right cert ok 5 s2 proxy listener should be up and have right cert not ok 6 ingress-gateway should have healthy endpoints for s1 not ok 7 s1 proxy should have been configured with max_connections in services ok 8 ingress-gateway should have healthy endpoints for s2 ```
2022-10-04 15:35:19 +00:00
echo "SKIPPING CASE: $SKIP_CASE"
return 0
fi
# Wipe state
wipe_volumes
# Push the state to the shared docker volume (note this is because CircleCI
# can't use shared volumes)
docker cp workdir/. envoy_workdir_1:/workdir
start_consul primary
if is_set $REQUIRE_SECONDARY; then
start_consul secondary
fi
2021-11-12 21:45:50 +00:00
if is_set $REQUIRE_PARTITIONS; then
2021-12-03 07:14:50 +00:00
docker_consul "primary" consul partition create -name ap1 > /dev/null
2021-11-12 21:45:50 +00:00
start_partitioned_client ap1
fi
if is_set $REQUIRE_PEERS; then
start_consul alpha
fi
echo "Setting up the primary datacenter"
pre_service_setup primary
if is_set $REQUIRE_SECONDARY; then
echo "Setting up the secondary datacenter"
pre_service_setup secondary
fi
2021-11-12 21:45:50 +00:00
if is_set $REQUIRE_PARTITIONS; then
echo "Setting up the non-default partition"
pre_service_setup ap1
fi
if is_set $REQUIRE_PEERS; then
echo "Setting up the alpha peer"
pre_service_setup alpha
fi
echo "Starting services"
start_services
# Run the verify container and report on the output
2021-11-12 21:45:50 +00:00
echo "Verifying the primary datacenter"
verify primary
if is_set $REQUIRE_SECONDARY; then
2021-11-12 21:45:50 +00:00
echo "Verifying the secondary datacenter"
verify secondary
fi
if is_set $REQUIRE_PEERS; then
echo "Verifying the alpha peer"
verify alpha
fi
}
function test_teardown {
init_vars
stop_services
}
function workdir_cleanup {
docker_kill_rm workdir
docker volume rm -f envoy_workdir &>/dev/null || true
}
function suite_setup {
# Cleanup from any previous unclean runs.
suite_teardown
docker network create envoy-tests &>/dev/null
# Start the volume container
#
# This is a dummy container that we use to create volume and keep it
# accessible while other containers are down.
docker volume create envoy_workdir &>/dev/null
docker run -d --name envoy_workdir_1 \
$WORKDIR_SNIPPET \
--net=none \
k8s.gcr.io/pause &>/dev/null
# TODO(rb): switch back to "${HASHICORP_DOCKER_PROXY}/google/pause" once that is cached
# pre-build the verify container
echo "Rebuilding 'bats-verify' image..."
2022-08-08 16:02:17 +00:00
retry_default docker build -t bats-verify -f Dockerfile-bats .
# if this fails on CircleCI your first thing to try would be to upgrade
# the machine image to the latest version using this listing:
#
# https://circleci.com/docs/2.0/configuration-reference/#available-linux-machine-images
echo "Checking bats image..."
docker run --rm -t bats-verify -v
# pre-build the consul+envoy container
echo "Rebuilding 'consul-dev-envoy:${ENVOY_VERSION}' image..."
2022-08-08 16:02:17 +00:00
retry_default docker build -t consul-dev-envoy:${ENVOY_VERSION} \
--build-arg ENVOY_VERSION=${ENVOY_VERSION} \
-f Dockerfile-consul-envoy .
# pre-build the test-sds-server container
echo "Rebuilding 'test-sds-server' image..."
2022-08-08 16:02:17 +00:00
retry_default docker build -t test-sds-server -f Dockerfile-test-sds-server test-sds-server
}
function suite_teardown {
docker_kill_rm verify-primary verify-secondary verify-alpha
# this is some hilarious magic
docker_kill_rm $(grep "^function run_container_" $self_name | \
sed 's/^function run_container_\(.*\) {/\1/g')
docker_kill_rm consul-primary consul-primary-server consul-secondary consul-secondary-server consul-ap1 consul-alpha consul-alpha-server
if docker network inspect envoy-tests &>/dev/null ; then
echo -n "Deleting network 'envoy-tests'..."
docker network rm envoy-tests
echo "done"
fi
workdir_cleanup
}
function run_containers {
for name in $@ ; do
run_container $name
done
}
function run_container {
docker_kill_rm "$1"
"run_container_$1"
}
function common_run_container_service {
local service="$1"
2021-11-12 21:45:50 +00:00
local CLUSTER="$2"
local httpPort="$3"
local grpcPort="$4"
docker run -d --name $(container_name_prev) \
-e "FORTIO_NAME=${service}" \
2021-11-12 21:45:50 +00:00
$(network_snippet $CLUSTER) \
"${HASHICORP_DOCKER_PROXY}/fortio/fortio" \
server \
-http-port ":$httpPort" \
-grpc-port ":$grpcPort" \
-redirect-port disabled >/dev/null
}
function run_container_s1 {
common_run_container_service s1 primary 8080 8079
}
2021-11-12 21:45:50 +00:00
function run_container_s1-ap1 {
common_run_container_service s1 ap1 8080 8079
}
function run_container_s2 {
common_run_container_service s2 primary 8181 8179
}
function run_container_s2-v1 {
common_run_container_service s2-v1 primary 8182 8178
}
function run_container_s2-v2 {
common_run_container_service s2-v2 primary 8183 8177
}
function run_container_s3 {
common_run_container_service s3 primary 8282 8279
}
function run_container_s3-v1 {
common_run_container_service s3-v1 primary 8283 8278
}
function run_container_s3-v2 {
common_run_container_service s3-v2 primary 8284 8277
}
function run_container_s3-alt {
common_run_container_service s3-alt primary 8286 8280
}
function run_container_s4 {
common_run_container_service s4 primary 8382 8281
}
function run_container_s1-secondary {
common_run_container_service s1-secondary secondary 8080 8079
}
function run_container_s2-secondary {
common_run_container_service s2-secondary secondary 8181 8179
}
function run_container_s2-ap1 {
common_run_container_service s2 ap1 8480 8479
}
function run_container_s3-ap1 {
common_run_container_service s3 ap1 8580 8579
}
function run_container_s1-alpha {
common_run_container_service s1-alpha alpha 8080 8079
}
function run_container_s2-alpha {
common_run_container_service s2-alpha alpha 8181 8179
}
function run_container_s3-alpha {
common_run_container_service s3-alpha alpha 8282 8279
}
function common_run_container_sidecar_proxy {
local service="$1"
2021-11-12 21:45:50 +00:00
local CLUSTER="$2"
# 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?
docker run -d --name $(container_name_prev) \
$WORKDIR_SNIPPET \
2021-11-12 21:45:50 +00:00
$(network_snippet $CLUSTER) \
$(aws_snippet) \
"${HASHICORP_DOCKER_PROXY}/envoyproxy/envoy:v${ENVOY_VERSION}" \
envoy \
2021-11-12 21:45:50 +00:00
-c /workdir/${CLUSTER}/envoy/${service}-bootstrap.json \
-l trace \
--disable-hot-restart \
--drain-time-s 1 >/dev/null
}
function run_container_s1-sidecar-proxy {
common_run_container_sidecar_proxy s1 primary
}
2021-11-12 21:45:50 +00:00
function run_container_s1-ap1-sidecar-proxy {
common_run_container_sidecar_proxy s1 ap1
}
function run_container_s1-sidecar-proxy-consul-exec {
docker run -d --name $(container_name) \
$(network_snippet primary) \
consul-dev-envoy:${ENVOY_VERSION} \
consul connect envoy -sidecar-for s1 \
-envoy-version ${ENVOY_VERSION} \
-- \
-l trace >/dev/null
}
function run_container_s2-sidecar-proxy {
common_run_container_sidecar_proxy s2 primary
}
function run_container_s2-v1-sidecar-proxy {
common_run_container_sidecar_proxy s2-v1 primary
}
function run_container_s2-v2-sidecar-proxy {
common_run_container_sidecar_proxy s2-v2 primary
}
function run_container_s3-sidecar-proxy {
common_run_container_sidecar_proxy s3 primary
}
function run_container_s3-v1-sidecar-proxy {
common_run_container_sidecar_proxy s3-v1 primary
}
function run_container_s3-v2-sidecar-proxy {
common_run_container_sidecar_proxy s3-v2 primary
}
function run_container_s3-alt-sidecar-proxy {
common_run_container_sidecar_proxy s3-alt primary
}
function run_container_s1-sidecar-proxy-secondary {
common_run_container_sidecar_proxy s1 secondary
}
function run_container_s2-sidecar-proxy-secondary {
common_run_container_sidecar_proxy s2 secondary
}
function run_container_s2-ap1-sidecar-proxy {
common_run_container_sidecar_proxy s2 ap1
}
function run_container_s3-ap1-sidecar-proxy {
common_run_container_sidecar_proxy s3 ap1
}
function run_container_s1-sidecar-proxy-alpha {
common_run_container_sidecar_proxy s1 alpha
}
function run_container_s2-sidecar-proxy-alpha {
common_run_container_sidecar_proxy s2 alpha
}
function run_container_s3-sidecar-proxy-alpha {
common_run_container_sidecar_proxy s3 alpha
}
function common_run_container_gateway {
local name="$1"
local DC="$2"
# 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?
docker run -d --name $(container_name_prev) \
$WORKDIR_SNIPPET \
$(network_snippet $DC) \
$(aws_snippet) \
"${HASHICORP_DOCKER_PROXY}/envoyproxy/envoy:v${ENVOY_VERSION}" \
envoy \
-c /workdir/${DC}/envoy/${name}-bootstrap.json \
-l trace \
--disable-hot-restart \
--drain-time-s 1 >/dev/null
}
function run_container_gateway-primary {
common_run_container_gateway mesh-gateway primary
}
function run_container_gateway-secondary {
common_run_container_gateway mesh-gateway secondary
}
function run_container_gateway-alpha {
common_run_container_gateway mesh-gateway alpha
}
function run_container_ingress-gateway-primary {
common_run_container_gateway ingress-gateway primary
}
function run_container_terminating-gateway-primary {
common_run_container_gateway terminating-gateway primary
}
function run_container_fake-statsd {
# This magic SYSTEM incantation is needed since Envoy doesn't add newlines and so
# we need each packet to be passed to echo to add a new line before
# appending.
docker run -d --name $(container_name) \
$WORKDIR_SNIPPET \
$(network_snippet primary) \
"${HASHICORP_DOCKER_PROXY}/alpine/socat:1.7.3.4-r1" \
-u UDP-RECVFROM:8125,fork,reuseaddr \
SYSTEM:'xargs -0 echo >> /workdir/primary/statsd/statsd.log'
}
function run_container_zipkin {
docker run -d --name $(container_name) \
$WORKDIR_SNIPPET \
$(network_snippet primary) \
"${HASHICORP_DOCKER_PROXY}/openzipkin/zipkin"
}
function run_container_jaeger {
docker run -d --name $(container_name) \
$WORKDIR_SNIPPET \
$(network_snippet primary) \
"${HASHICORP_DOCKER_PROXY}/jaegertracing/all-in-one:1.11" \
--collector.zipkin.http-port=9411
}
function run_container_test-sds-server {
docker run -d --name $(container_name) \
$WORKDIR_SNIPPET \
$(network_snippet primary) \
"test-sds-server"
}
function container_name {
echo "envoy_${FUNCNAME[1]/#run_container_/}_1"
}
function container_name_prev {
echo "envoy_${FUNCNAME[2]/#run_container_/}_1"
}
# This is a debugging tool. Run via './run-tests.sh debug_dump_volumes'
function debug_dump_volumes {
docker run --rm -it \
$WORKDIR_SNIPPET \
-v ./:/cwd \
--net=none \
"${HASHICORP_DOCKER_PROXY}/alpine" \
cp -r /workdir/. /cwd/workdir/
}
function run_container_tcpdump-primary {
# To use add "tcpdump-primary" to REQUIRED_SERVICES
common_run_container_tcpdump primary
}
function run_container_tcpdump-secondary {
# To use add "tcpdump-secondary" to REQUIRED_SERVICES
common_run_container_tcpdump secondary
}
function run_container_tcpdump-alpha {
# To use add "tcpdump-alpha" to REQUIRED_SERVICES
common_run_container_tcpdump alpha
}
function common_run_container_tcpdump {
local DC="$1"
# we cant run this in circle but its only here to temporarily enable.
2022-08-08 16:02:17 +00:00
retry_default docker build -t envoy-tcpdump -f Dockerfile-tcpdump .
docker run -d --name $(container_name_prev) \
$(network_snippet $DC) \
-v $(pwd)/workdir/${DC}/envoy/:/data \
--privileged \
envoy-tcpdump \
-v -i any \
-w "/data/${DC}.pcap"
}
case "${1-}" in
"")
echo "command required"
exit 1 ;;
*)
"$@" ;;
esac