How to fix : Received RST_STREAM with error code 2 when using nginx reverse proxy
Asked Answered
P

1

6

I'm currently using the dialogflow api on a raspberry. Everything works fine when calling StreamingDetectIntent method using grpc. I have to use multiples apis on my product and so, I'm trying to put a reverse proxy in front of them. Like that, I can call only one address I'm using nginx to reverse proxy my GRPC request to google api. I have no problem when calling simple method, but when calling a streaming method like StreamingDetectIntent, I got an error during the request.

Dialogflow do not have problem to get the audio flux coming from my client, but I got problem to get the last part of the request, the downstream flux.

Here is the error that my client give me :

grpc._channel._Rendezvous: <_Rendezvous of RPC that terminated with:
        status = StatusCode.INTERNAL
        details = "Received RST_STREAM with error code 2"
        debug_error_string = "{"created":"@1567173815.816362297","description":"Error received from peer ipv4:163.172.143.250:443","file":"src/core/lib/surface/call.cc","file_line":1041,"grpc_message":"Received RST_STREAM with error code 2","grpc_status":13}"
>

and here the error I can see in Nginx log :

upstream sent frame for closed stream 1 while reading upstream, client: ..., server: exemple.com, request: "POST /google.cloud.dialogflow.v2beta1.Sessions/StreamingDetectIntent HTTP/2.0", upstream: "grpcs://...:443", host: "example.com:443"

I've tried to increase grpc_buffer_size parameter to big value, but didn't worked..

Here is my current Nginx config :

user  nginx;
worker_processes  1;

error_log  /var/log/nginx/error.log debug;
pid        /var/run/nginx.pid;

events {
    worker_connections  1024;
}

http {

    include       /etc/nginx/mime.types;
    default_type  application/octet-stream;

    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                      '$status $body_bytes_sent "$http_referer" '
                      '"$http_user_agent" "$http_x_forwarded_for"';

    access_log  /var/log/nginx/access.log  main;

    sendfile        on;

    keepalive_timeout  65;

    client_max_body_size 4000M;
    grpc_read_timeout 1d;
    grpc_send_timeout 1d;
    # this seems to fix it; but see comment in README.md
    grpc_buffer_size 100M;

    include /etc/nginx/conf.d/*.conf;

    server {

        # SSL configuration
        listen 443 ssl http2;

        access_log /var/log/nginx/access_grpc.log main;

        location / {
            grpc_pass grpcs://dialogflow.googleapis.com:443;
        }

        ssl_certificate /etc/letsencrypt/live/example.com/fullchain.pem;
        ssl_certificate_key/etc/letsencrypt/live/exemple.com/privkey.pem;
    }

    server {

        if ($host = example.com) {
            return 301 https://$host$request_uri;
        }

        listen 80 ;
        listen [::]:80 ;

        return 404; # managed by Certbot

    }

}

Plummet answered 30/8, 2019 at 14:31 Comment(2)
Found a solution? I’ve got the same problem with Nginx 1.18Fabliau
Have you found a solution?Praetorian
H
3

There may be a problem with GRPC Metadata (especially with user-agent). Take a look at these issues:

https://github.com/grpc/grpc-go/issues/1888

https://github.com/improbable-eng/grpc-web/issues/145

Halstead answered 19/12, 2019 at 17:6 Comment(0)

© 2022 - 2024 — McMap. All rights reserved.