I have an HTTP server that when it recieves a request calls on an underlying gRPC server.
I have chosen to abstract away the gRPC call with an interface, to make testing of the http server easier.
The problem is that I am constantly getting the errors:
rpc error: code = Canceled desc = grpc: the client connection is closing
or
rpc error: code = Canceled desc = context canceled
And as I understand both of these are related to the context getting passed into the grpc call. And that I want the context to be alive throughout both the HTTP and gRPC calls.
type SetterGetter interface {
Getter(key string) (val string)
}
type Service struct {
sg SetterGetter
ctx context.Context
}
func (s *Service) getHandler(rw http.ResponseWriter, r *http.Request) {
key := r.URL.Query()["key"][0]
res := s.sg.Getter(key)
fmt.Fprintf(rw, "Successfully got value: %s\n", res)
}
func main() {
s := new(Service)
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
s.sg = gc.NewClientwrapper(ctx)
http.HandleFunc("/get", s.getHandler)
log.Fatal(http.ListenAndServe(port, nil))
}
And my Getter implementation looks like this:
type clientwrapper struct {
sc pb.ServicesClient
ctx context.Context
}
func NewClientwrapper(ctx context.Context) *clientwrapper {
cw := new(clientwrapper)
conn, err := grpc.Dial(address, grpc.WithInsecure(), grpc.WithBlock())
if err != nil {
err = fmt.Errorf("Error could not dial address: %v", err)
}
defer conn.Close()
cw.ctx = ctx
cw.sc = pb.NewServicesClient(conn)
return cw
}
func (cw *clientwrapper) Getter(key string) (val string) {
// Make the GRPC request
res, err := cw.sc.Get(cw.ctx, &pb.GetRequest{Key: key})
if err != nil {
return ""
}
getVal := res.GetValue()
return getVal
}
So here I am creating a context in my http servers main menu, and passing it onwards. I do it like this because it worked if I removed my interface and put everything in the main file.
I have also tried to create the context both in the http handler and passing it to the Getter and I have also tried creating it in the Getter itself.
I think the correct approach is to create the context in the http request using the context that gets created by the request and then passing it to the grpc Getter. Like such:
func (s *Service) getHandler(rw http.ResponseWriter, r *http.Request) {
// Create it like such
ctx, cancel := context.WithTimeout(r.Context(), 100*time.Second)
key := r.URL.Query()["key"][0]
// And pass it onwards (of course we need to change function signature for this to work)
res := s.sg.Getter(ctx, key)
fmt.Fprintf(rw, "Successfully got value: %s\n", res)
}
So how should I create my context here, to not get these errors?
If your goal is to keep a long-running task running in the background, that doesn't cancel when the request is finalized, then don't use the request's context. Use context.Background() instead.
For example:
func (s *Service) getHandler(rw http.ResponseWriter, r *http.Request) {
ctx, cancel := context.WithTimeout(context.Background(), 100*time.Second)
// ...
Related
I am writing an HTTP server in Go, which uses the following pattern to handle API output:
func handler(w http.ResponsWriter, r *http.Request) {
defer reply(w, r, L)() //L is a Logger
//do things...
}
func reply(w http.ResponseWriter, r *http.Request, log Logger) func() {
cid := []byte{0, 0, 0, 0}
if log != nil {
rand.Read(cid)
log.Debug("[%x] %s %s", cid, r.Method, r.URL.String())
}
entry := time.Now()
return func() {
if log != nil {
defer log.Debug("[%x] elapsed %d millis", cid, time.Since(entry).Milliseconds())
}
_, err := w.Write(nil)
if err == http.ErrHijacked {
return //API is a WEBSOCKET entry point, do nothing
}
//handle common output logic for normal HTTP APIs...
}
}
The reason I do this, is that I found this comment in the standard library:
// ErrHijacked is returned by ResponseWriter.Write calls when
// the underlying connection has been hijacked using the
// Hijacker interface. A zero-byte write on a hijacked
// connection will return ErrHijacked without any other side
// effects.
ErrHijacked = errors.New("http: connection has been hijacked")
However following the Write() method, I got this comment:
// Write writes the data to the connection as part of an HTTP reply.
//
// If WriteHeader has not yet been called, Write calls
// WriteHeader(http.StatusOK) before writing the data. If the Header
// does not contain a Content-Type line, Write adds a Content-Type set
// to the result of passing the initial 512 bytes of written data to
// ...
Write([]byte) (int, error)
My questions are:
Is it OK to use my code to safely detect if a HTTP connection is hijacked? I only want to check the connection is hijacked or not, but do NOT want it to add headers for me!
Since the ResponseWriter is an interface, I cannot click through the source code to find out how the standard library implements that method. In general, how can I drill down to the standard library (or any open source code) to find out the implementation of an interface?
Thanks to Cerise, I found the source code of the standard response.Writer:
func (w *response) write(lenData int, dataB []byte, dataS string) (n int, err error) {
if w.conn.hijacked() {
if lenData > 0 {
caller := relevantCaller()
w.conn.server.logf("http: response.Write on hijacked connection from %s (%s:%d)", caller.Function, path.Base(caller.File), caller.Line)
}
return 0, ErrHijacked
}
... ....
So, as said in the document, there is NO side effect.
could anyone help me here please as I'm new to golang? I have a yaml file which looks like this:
port: 5000
handlers:
- name: test1
uri: /api/test1
response:
status: 200
body: test1
- name: test2
uri: /api/test2
response:
status: 500
body: test2
based on this file I want to create a server. Currently I'm trying to do it this way, but looks like it doesn't work as expected.
What am I doing wrong and what is the better way to achieve what I need?
package main
import (
"fmt"
"io/ioutil"
"log"
"net/http"
"gopkg.in/yaml.v2"
)
func main() {
config := parseYaml("conf.yaml")
configHandlers := config.Handlers
mux := http.NewServeMux()
for _, handler := range *configHandlers {
mux.HandleFunc(*handler.Uri, func(w http.ResponseWriter, r *http.Request) {
w.WriteHeader(*handler.Response.Status)
fmt.Fprintf(w, *handler.Response.Body)
})
}
log.Fatal(http.ListenAndServe(fmt.Sprintf(":%s", *config.Port), mux))
}
type YamlConfig struct {
Port *string `yaml:"port"`
Handlers *[]HandlerConfig `yaml:"handlers"`
}
type HandlerConfig struct {
Uri *string `yaml:"uri"`
Name *string `yaml:"name"`
Response *Response `yaml:"response"`
}
type Response struct {
Status *int `yaml:"status"`
Body *string `yaml:"body"`
}
func (c *YamlConfig) parseYaml(data []byte) error {
return yaml.Unmarshal(data, c)
}
func parseYaml(path string) YamlConfig {
data, err := ioutil.ReadFile(path)
if err != nil {
log.Fatal(err)
}
var config YamlConfig
if err := config.parseYaml(data); err != nil {
log.Fatal(err)
}
return config
}
Update:
If I run this server then regardless of which endpoint I hit, it will always return me 500 and test2 in body
What you're seeing is seemingly a common pitfall for people:
configHandlers := config.Handlers
mux := http.NewServeMux()
for _, handler := range *configHandlers {
mux.HandleFunc(*handler.Uri, func(w http.ResponseWriter, r *http.Request) {
w.WriteHeader(*handler.Response.Status)
fmt.Fprintf(w, *handler.Response.Body)
})
}
The for loop, on each iteration, reassigns the handler variable. In the loop body, you create a new function and pass it to mux.HandlerFun. These function bodies kind of inherit the outer scope, and access this handler variable. The variable is reassigned outside of the functions, and thus the values each handler function has access to changes with it. What you can do to address the issue is mask the handler variable the loop uses, and create a scope that is unique to each handler. The classic way in languages like JavaScript (where this is - or used to be back when I wrote some JS - a common issue) is to wrap the code in an IIFE (Immediately Invoked Function Expression):
for _, handler := range *configHandlers {
func (handler *HandlerConfig) { // handler is now the argument passed to this function
mux.HandleFunc(*handler.Uri, func(w http.ResponseWriter, r *http.Request) {
w.WriteHeader(*handler.Response.Status)
fmt.Fprintf(w, *handler.Response.Body)
})
}(handler) // call the function with the _current_ value of handler
}
This is a tad messy, and because golang is properly block-scoped, you can just do this:
for _, handler := range *configHandlers {
h := handler // create a variable in the inner scope
mux.HandleFunc(*handler.Uri, func(w http.ResponseWriter, r *http.Request) {
// now h will reference a copy unique to each iteration
w.WriteHeader(*h.Response.Status)
fmt.Fprintf(w, *h.Response.Body)
})
}
That ought to fix it. I've noticed some weirdness with your use of pointers in the types you've added to your question, though... Fields like Port being of type *string? Why wouldn't you just use string? No Same for the Body and Status fields in the Response type. By changing them to plain string fields you don't have to dereference them in your handler functions. It will look a lot cleaner.
A bigger worry is this field:
Handlers *[]HandlerConfig `yaml:"handlers"`
I'm not sure if you really know what the type of this field is, but it makes next to no sense. Handlers is now a pointer to a slice of HandlerConfig values. I'm assuming you wanted this field to be:
// Handlers is a slice of HandlerConfig values:
Handlers []HandlerConfig `yaml:"handlers"`
// or Handlers is a slice of pointers to HandlerConfig values
Handlers []*HandlerConfig `yaml:"handlers"`
Generally speaking, a pointer to a slice, especially in a config type is bad code.
If you define a struct that will represent the configuration in your YAML file, you can unmarshall the yaml into an instantiated struct of that type using the yaml package. From there, you can reference the fields in the struct as any other struct.
package main
import (
"fmt"
"gopkg.in/yaml.v2"
)
type YamlExample struct {
FieldOne string `yaml:"fieldOne"`
NestedField struct {
Name string `yaml:"name"`
} `yaml:"nestedField"`
}
const YamlEx string = `
fieldOne: one
nestedField:
name: nestedFieldName
`
func main() {
var yamlE YamlExample
err := yaml.Unmarshal([]byte(YamlEx), &yamlE)
if err != nil {
panic(err)
}
fmt.Printf("%+v\n", yamlE)
}
Link to example.
In your case, you'd probably want to handle the routes in a struct and then reference the fields in the struct for things like route name, how to handle the body of the request, etc. If your YAML is stored in a file, you'll have to use something like the io package to read the file into a byte array that the YAML package can parse. See here for a reference.
I am learning Go and have a quick question about http handler implementation in Go.
I am asking it in a small sample code.
So assume there is a handler function called Test() as defined like below
func Test() func(http.ResponseWriter, *http.Request) {
return func(w http.ResponseWriter, r *http.Request) {
params := r.URL.Query()
name := params.Get("name")
if name == "axy" {
common.UpdateHttpResponse("Trying to updating the response", w, http.StatusBadRequest)
//return
}
w.WriteHeader(http.StatusOK)
}
}
lets say the UpdateHttpResponse method is also define in common package as below:
func LogExtractionErrorResponse(errMsg string, w http.ResponseWriter, responseHeader int) {
fmt.Printf("%s", errMsg)
jsonErrorOut := map[string]string{
"Error": errMsg,
}
w.WriteHeader(responseHeader)
encodedResponse, _ := json.Marshal(jsonErrorOut)
if w != nil {
w.Write(encodedResponse)
}
}
I call the Test() HTTP handler in the http server part as below.
// this is how Test() http handler is called as well
http.HandleFunc("/test", httpserver.Test())
So here is my question:
Based on my understanding all values are passed by value in go (as discussed in this thread as well)
In that case why if the http handler is called with a parameter (i.e localhost:PORT_NUM/test?name=axy), i observe "StatusBadRequest" in resonse. In other words, why the commented "return" keyword is not needed and why the header response is not overwritten by "w.WriteHeader(http.StatusOK)" at the end of Test() http handler?
I have a basic HTTP server that accepts a request and returns data from a data store.
Each HTTP request does the following things:
Create a context with timeout
Create a read request (custom type)
Push read request onto channel
Wait for response and serve data
Here's the basic pseudo code:
package main
import (
"context"
"net/http"
"time"
)
type dataRequest struct {
data chan string
ctx context.Context
}
func handler(reqStream chan dataRequest) http.HandlerFunc {
return func(w http.ResponseWriter, r *http.Request) {
ctx, cancel := context.WithTimeout(r.Context(), 5*time.Second)
defer cancel()
req := dataRequest{
data: make(chan string),
ctx: ctx,
}
select {
case reqStream <- req:
// request pushed to que
case <-ctx.Done():
// don't push onto reqStream if ctx done
}
select {
case <-ctx.Done():
// don't try and serve content if ctx done
case data := <-req.data:
// return data to client
}
}
}
func main() {
dataReqs := make(chan dataRequest)
go func() {
for {
select {
case req := <-dataReqs:
select {
case <-req.ctx.Done():
// don't push onto data channel if ctx done
case req.data <- "some data":
// get data from store
}
}
}
}()
http.HandleFunc("/", handler(dataReqs))
http.ListenAndServe(":8080", nil)
}
My question is, because the context could finish at any time due to the deadline being exceeded or the client cancelling the request, is my current approach correct for handling this in multiple places or is there a more elegant solution?
seems to me that it'll work.
few comments -
you can return in the first case of <- ctx.Done()
you're already waiting for req.ctx.Done() in the data store handler so you can completely remove the first select {} statement and just publish to the data requests channel. not sure about performance hits for the rare cases when the context is done so early before the request is even published...
I need to make a request to a server that return different responses at different times. I mean, the server generate different responses and these responses take different execution time, so server return the responses as soon as they are available.
And I want print in the screen (by the moment, I'd settle with that) these responses as soon as the server returns me.
All what I could do until now is print the responses but only when the server returns all the responses. So if the first response take 1sec, and the last response take 10sec, my code needs to wait 10sec to print all the messages.
EDIT: to add code I have:
//Config is gotten from yml file
RestConfig = Config["rest"].(map[string]interface{})
ServerConfig = Config["server"].(map[string]interface{})
RequestUrl := ServerConfig["url"]
RequestReader := bytes.NewReader(body)
Request, _ := http.NewRequest("POST", RequestUrl.(string), RequestReader)
//AppendHeaders append the needing headers to the request
client.AppendHeaders(Request, RestConfig["headers"])
//the type of client.HttpClient is *http.Client
Response, _ := client.HttpClient.Do(Request)
//And to print in the screen
defer Response.Body.Close()
fmt.Println( "-> Receiving response:\n---\n" )
fmt.Println( Response , "\n---\n-> Response body:\n---\n")
body_resp, _ := ioutil.ReadAll(Response.Body)
fmt.Println( string(body_resp) )
fmt.Println( "\n--\n")
Any way to do it??
Thank you very much.
Finally my code , is like this:
package main
import (
"fmt"
"log"
"bytes"
"strings"
"bufio"
"net/http"
)
func main() {
var body = "The body"
RequestReader := bytes.NewReader([]byte(body))
req, err := http.NewRequest("POST", "the_url", RequestReader)
if err != nil {
log.Fatal(err)
}
req.Header.Add("Accept", "application/xml")
req.Header.Add("Content-Type", "application/xml")
req.Header.Add("AG-Authorization", "key")
req.Header.Add("AG-Forwarded-Hosts", "*")
resp, err := (&http.Client{}).Do(req)
if err != nil {
log.Fatal(err)
}
reader := bufio.NewReader(resp.Body)
message := ""
for {
line, err := reader.ReadBytes('\n')
if err != nil {
log.Fatal(err)
}
message = message + string(line)
if strings.Contains(message, "<!-- End mark for each message -->"){
fmt.Println(message)
message = ""
}
}
}
Thank everyone.
The context package is what you are looking for.
The context package is responsible for signal cancelation and operation deadlines for processes and server requests. This has two public methods: WithCancel and WithTimeout. The Context associated with an incoming request is typically canceled when the request handler returns.
For your specific case you can use the WithTimeout method for setting a deadline on requests to backend servers.
// WithTimeout returns a copy of parent whose Done channel is closed as soon as
// parent.Done is closed, cancel is called, or timeout elapses. The new
// Context's Deadline is the sooner of now+timeout and the parent's deadline, if
// any. If the timer is still running, the cancel function releases its
// resources.
func WithTimeout(parent Context, timeout time.Duration) (Context, CancelFunc)
And here is a snippet taken from https://blog.golang.org/context/server/server.go
timeout, err := time.ParseDuration(req.FormValue("timeout")) // set a time limit in your post
if err == nil {
// The request has a timeout, so create a context that is
// canceled automatically when the timeout expires.
ctx, cancel = context.WithTimeout(context.Background(), timeout)
} else {
ctx, cancel = context.WithCancel(context.Background())
}
defer cancel() // Cancel ctx as soon as handleSearch returns.
For further reading take a look at this article:
https://blog.golang.org/context