mirror of
https://github.com/andwati/zitadel.git
synced 2026-10-19 14:12:52 +00:00
5934e07960
# Which Problems Are Solved Panics may cause a service discruption by unexpectedly closing an request's connection. Or in the case of gRPC completely killing the service. Allthough panics are still individual bugs that need to be solved, this PR makes sure a panic is gracefully handled and an understandable error is returned to the client. # How the Problems Are Solved - Recover in the middleware interceptors for the 3 API protocols (HTTP, gRPC, connect). - HTTP middleware uses formatted responses for: - OIDC errors (JSON formatted response) - UI (error page rendering) - SCIM - Upon recovery an alert level log is printed (ERROR+4 for stdlib log handlers) # Additional Context - internal observation --------- Co-authored-by: Livio Spring <[email protected]> Co-authored-by: Livio Spring <[email protected]>
83 lines
2.7 KiB
Go
83 lines
2.7 KiB
Go
package middleware
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"net/http"
|
|
|
|
"github.com/zitadel/sloggcp"
|
|
|
|
"github.com/zitadel/zitadel/backend/v3/instrumentation/logging"
|
|
"github.com/zitadel/zitadel/internal/zerrors"
|
|
)
|
|
|
|
// RecoverHandler recovers from panics in the HTTP handler chain
|
|
// and calls the provided writeResponse function to write an appropriate response to the client.
|
|
//
|
|
// The request context is canceled with the panic error as the cause.
|
|
func RecoverHandler(writeResponse func(w http.ResponseWriter, r *http.Request, err error)) func(http.Handler) http.Handler {
|
|
return func(next http.Handler) http.Handler {
|
|
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
|
ctx, cancel := context.WithCancelCause(r.Context())
|
|
r = r.WithContext(ctx)
|
|
|
|
defer func() {
|
|
var err error
|
|
if rec := recover(); rec != nil {
|
|
recErr, ok := rec.(error)
|
|
if !ok {
|
|
recErr = fmt.Errorf("%v", rec)
|
|
}
|
|
err = zerrors.ThrowInternal(recErr, zerrors.IDRecover, "Errors.Internal")
|
|
logRecovered(ctx, err)
|
|
writeResponse(w, r, err)
|
|
}
|
|
cancel(err)
|
|
}()
|
|
|
|
next.ServeHTTP(w, r)
|
|
})
|
|
}
|
|
}
|
|
|
|
// FallbackRecoverHandler recovers from panics in the HTTP handler chain
|
|
// and returns a 500 Internal Server Error response.
|
|
// The request context is canceled with the panic error as the cause,
|
|
// so that any ongoing operations can be stopped and cleaned up.
|
|
//
|
|
// The response is sent as a text/plain response.
|
|
// It is used as a last line of defense to prevent the server from crashing
|
|
// due to panics in the handlers.
|
|
// Protocols (OIDC, SAML, HTML etc.) should use [RecoverHandler] to write
|
|
// properly formatted error responses to the clients.
|
|
func FallbackRecoverHandler() func(http.Handler) http.Handler {
|
|
return RecoverHandler(func(w http.ResponseWriter, r *http.Request, _ error) {
|
|
http.Error(w, http.StatusText(http.StatusInternalServerError), http.StatusInternalServerError)
|
|
})
|
|
}
|
|
|
|
// RecoverHandlerWithError is similar to [RecoverHandler] but returns an error instead of writing a response directly.
|
|
func RecoverHandlerWithError(next HandlerFuncWithError) HandlerFuncWithError {
|
|
return func(w http.ResponseWriter, r *http.Request) (err error) {
|
|
ctx, cancel := context.WithCancelCause(r.Context())
|
|
r = r.WithContext(ctx)
|
|
defer func() {
|
|
if rec := recover(); rec != nil {
|
|
recErr, ok := rec.(error)
|
|
if !ok {
|
|
recErr = fmt.Errorf("%v", rec)
|
|
}
|
|
err = zerrors.ThrowInternal(recErr, zerrors.IDRecover, "Errors.Internal")
|
|
logRecovered(ctx, err)
|
|
}
|
|
cancel(err)
|
|
}()
|
|
return next(w, r)
|
|
}
|
|
}
|
|
|
|
func logRecovered(ctx context.Context, err error) {
|
|
logger := logging.FromCtx(ctx)
|
|
logger.Log(ctx, sloggcp.LevelAlert, "recovered from panic", "err", err)
|
|
}
|