我遵循这个tutorial,以便将grcp服务代码转换为超文本传输协议(它在Linux上运行)。特使API更新到了v3,所以我也遵循了这个example。我已经将v3接口的envoy_config更新了,所以它符合这个版本的新要求。
但是,当我部署docker镜像时,我在尝试访问API端点时遇到错误。例如,当我运行这个命令curl http://localhost:51051/greeting (它是生成的API的公开端点)时,我得到以下错误:upstream connect error or disconnect/reset before headers. reset reason: connection failure,状态码为503。
当我运行以下命令sudo docker ps时,我可以看到Port列中没有显示端口。因此,我尝试使用docker run -p ...或docker run --expose ...命令来映射/公开它们,但它们仍然没有出现。
我现在有点迷失了,因为我根本不知道docker,我甚至不知道我的问题是否与端口暴露有关。
但不管怎样,我不得不说grpc服务在java服务器/客户端上工作得很好。只是Web API (特使)不起作用。
如果你想要更多的资料,我已经上传了整个项目的here,所以你可以自己重现这个问题。
否则,这里是envoy-config.yml (公开的API端点是地址#2:0.0.0.0:51051):
admin:
access_log_path: /tmp/admin_access.log
address:
socket_address: { address: 0.0.0.0, port_value: 9901 } #1
static_resources:
listeners:
- name: main-listener
address:
socket_address: { address: 0.0.0.0, port_value: 51051 } #2
filter_chains:
- filters:
- name: envoy.filters.network.http_connection_manager
typed_config:
"@type": type.googleapis.com/envoy.extensions.filters.network.http_connection_manager.v3.HttpConnectionManager # Explicit v3
stat_prefix: grpc_json
codec_type: AUTO
route_config:
name: local_route
virtual_hosts:
- name: local_service
domains: ["*"]
routes:
- match: { prefix: "/" , grpc: {}} # 3a grpc:{} means that requests are only forwarded if they are found in the grpc service definition, returning 404 for others
route: { cluster: grpc-backend-services, timeout: { seconds: 60 } } #3b
http_filters:
- name: envoy.filters.http.grpc_json_transcoder
typed_config:
"@type": type.googleapis.com/envoy.extensions.filters.http.grpc_json_transcoder.v3.GrpcJsonTranscoder
proto_descriptor: "/data/greeting_service_definition.pb" #4
services: ["helloworld.HelloService"] #5
print_options:
add_whitespace: true
always_print_primitive_fields: true
always_print_enums_as_ints: false
preserve_proto_field_names: false #6
- name: envoy.filters.http.router
clusters:
- name: grpc-backend-services #7
connect_timeout: 1.25s
type: logical_dns
lb_policy: round_robin
dns_lookup_family: V4_ONLY
typed_extension_protocol_options:
envoy.extensions.upstreams.http.v3.HttpProtocolOptions:
"@type": type.googleapis.com/envoy.extensions.upstreams.http.v3.HttpProtocolOptions
explicit_http_config:
http2_protocol_options: {}
load_assignment:
cluster_name: grpc-backend-services
endpoints:
- lb_endpoints:
- endpoint:
address:
socket_address:
address: 127.0.0.1 #8
port_value: 53000下面是运行protoc和docker的脚本
#!/usr/bin/env bash
if ! [ -x "$(command -v protoc)" ] ; then
echo "you do not seem to have the protoc executable on your path"
echo "we need protoc to generate a service defintion (*.pb file) that envoy can understand"
echo "download the precompiled protoc executable and place it in somewhere in your systems PATH!"
echo "goto: https://github.com/protocolbuffers/protobuf/releases/latest"
echo "choose:"
echo " for linux: protoc-3.6.1-linux-x86_64.zip"
echo " for windows: protoc-3.6.1-win32.zip"
echo " for mac: protoc-3.6.1-osx-x86_64.zip"
exit 1
fi
# generate the greeting_service_definition.pb file that we can pass to envoy so that knows the grpc service
# we want to expose
protoc -I. -Isrc/main/proto --include_imports \
--include_source_info \
--descriptor_set_out=greeting_service_definition.pb \
src/main/proto/HelloService.proto
if ! [ $? -eq 0 ]; then
echo "protobuf compilation failed"
exit 1
fi
# now we can start envoy in a docker container and map the configuration and service definition inside
# we use --network="host" so that envoy can access the grpc service at localhost:<port>
# the envoy-config.yml has configured envoy to run at port 51051, so you can access the HTTP/JSON
# api at localhost:51051
if ! [ -x "$(command -v docker)" ] ; then
echo "docker command is not available, please install docker"
echo "Install docker: https://store.docker.com/search?offering=community&type=edition"
exit 1
fi
# check if sudo is required to run docker
if [ "$(groups | grep -c docker)" -gt "0" ]; then
echo "Envoy will run at port 51051 (see envoy-config.yml)"
docker run -it --rm --name envoy --network="host" \
-v "$(pwd)/greeting_service_definition.pb:/data/greeting_service_definition.pb:ro" \
-v "$(pwd)/envoy-config.yml:/etc/envoy/envoy.yaml:ro" \
envoyproxy/envoy:v1.18.2
else
echo "you are not in the docker group, running with sudo"
echo "Envoy will run at port 51051 (see envoy-config.yml)"
sudo docker run -it --rm --name envoy --network="host"\
-v "$(pwd)/greeting_service_definition.pb:/data/greeting_service_definition.pb:ro" \
-v "$(pwd)/envoy-config.yml:/etc/envoy/envoy.yaml:ro" \
envoyproxy/envoy:v1.18.2
fi发布于 2021-05-26 03:05:35
503状态是因为代码转换操作失败-看起来特使无法连接到gRPC服务。
您已经将特使配置为在host.docker.internal:53000上连接到gRPC -如果您在linux上使用docker,而不是在Mac/Windows上使用docker-desktop,这可能是一个问题。
但是java服务器也监听8080端口,而不是53000端口,所以我认为你只需要改变这一点。
编辑
我确认了“docker”在Ubuntu18.04上可以和你的项目一起工作,就像这样调用host.docker.internal:docker run -it --rm --name envoy --add-host=host.docker.internal:host-bridge -p 51051:5105。在我测试的版本(f9e7c1)中,lb_endpoints内的端口需要从53000更改为8080。
注意,这是使用默认的“网桥”联网模式,因此需要端口映射以例如curl访问代码转换的代理端点。
你可以在“主机”模式下运行特使容器--这会简单一点,因为你可以通过本地主机(无端口映射) curl到特使的转码端点,而且特使也可以通过本地主机IP访问java服务器。
有关处理这类容器间依赖关系的方法的详细总结,请参阅this answer。
https://stackoverflow.com/questions/67691178
复制相似问题