Hi, I am have encountered issues on Android and iOS clients after proxying gRPC via HAProxy. From the clients we get UNKOWN status coming from Http2ClientStreamTransportState exceptions. The logs show
- "missing GRPC status in response”
- "UNKNOWN: HTTP status code 200 missing GRPC status, inferred error from HTTP status code”
- UNKNOWN: HTTP status code 200 invalid content-type: null …
For #3 we sometimes get headers and sometimes trailers logged. In both cases the data is clearly corrupted.
I get multiple entries for keys like “grpc-encoding”, “commerce-details-bin” with values that belong to other keys. furthermore I have values that are base64 encoded that are encoding “<life-feed><tab-delimiter>{status_code} {error_description}<line-terminator>{request-id}”
Most worrying is that in a single log I have multiple request-ids from other requests.
This started happening 8 hours after we started proxying routing some of the traffic to another backend. The backend has been used internally for weeks without issue. The traffic has been flowing through HAProxy at layer 7 for months without issue.
Originally the issue pointed to HAProxy but I don’t know how a proxy could possibly get 3 metadata values into a base64 value. That would be encoded only in the server.
Hi All, I’m looking for a good example of using AsyncAuthInterceptor in c#
i’m testing locally on a grpc server that doesn’t have ssl setup currently, and i can see that my asyncauthinterceptor never gets called, thus my authorization token is never applied to my header
```
var asyncAuthInterceptor = new AsyncAuthInterceptor(async (context, metadata) =>
{
await Task.Delay(100).ConfigureAwait(false); // make sure the operation is asynchronous.
metadata.Add("authorization", "SECRET_TOKEN");
Console.WriteLine(metadata.ToString());
});
_callCredentials = CallCredentials.FromInterceptor(asyncAuthInterceptor);
_channel = new Channel(grpcServerURL, ChannelCredentials.Insecure);
_client = new Publisher.PublisherClient(_channel);
…
private void Publish(PublishRequest request)
{
var callOptions = new CallOptions(credentials: _callCredentials);
var response = _client.Publish(request, callOptions);
Console.WriteLine($"Response: {response}");
}
`
users.v1.Users@127.0.0.1:50051> show service
+---------+----------+--------------+------------------+
| SERVICE | RPC | REQUEST TYPE | RESPONSE TYPE |
+---------+----------+--------------+------------------+
| Users | Register | NewUser | RegisterResponse |
+---------+----------+--------------+------------------+
Hello, new contributor here. I decided the best way to start contributing is to write unit tests here and there. This is on grpc-java. When I try to run unit tests however, I get the following error.
/home/aditya/grpc-java/context/src/main/java/io/grpc/Context.java:511: error: An unhandled exception was thrown by the Error Prone static analysis plugin.
break;Not sure why, anyone know?
I am working to build a middleware that has following capabilities
- intercepts all unary calls
- has ability to read the full request message
- has ability to generate response from interceptor and short circuit the response (skips the rest of the interceptors and service execution)
Any example pointers on how to implement 2 & 3, will be helpful.
Hello, I have implemented grpc.lb.v1.load_balancer.proto, however my client can't connect it.
To test load balancer, I am using a simple python client:
def create_channel(conn):
opts = [("grpc.lb_policy_name", "grpclb"), ("grpc.dns_enable_srv_queries", True)]
channel = grpc.insecure_channel(conn, opts)
return testservice_grpc.testStub(channel)
t = create_channel("ns.ourdomain.com")
t.Test(empty.Empty())However I am getting this error:
_Rendezvous: <_Rendezvous of RPC that terminated with:
status = StatusCode.UNAVAILABLE
details = "connections to all backends failing"
debug_error_string = "{"created":"@1569236325.444837337","description":"Failed to pick subchannel","file":"src/core/ext/filters/client_channel/client_channel.cc","file_line":3818,"referenced_errors":[{"created":"@1569236325.444770342","description":"connections to all backends failing","file":"src/core/ext/filters/client_channel/lb_policy/round_robin/round_robin.cc","file_line":328,"grpc_status":14}]}"As you can see from error, it is trying to use round robin even though our srv record's service name is grpclb. Our srv record for load balancer is below:
$ dig srv _grpclb._tcp.ourdomain.com
; <<>> DiG 9.14.6 <<>> srv _grpclb._tcp.ourdomain.com
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 50996
;; flags: qr rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 4, ADDITIONAL: 3
;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
; COOKIE: xxxxxx (good)
;; QUESTION SECTION:
;_grpclb._tcp.ourdomain.com. IN SRV
;; ANSWER SECTION:
_grpclb._tcp.ourdomain.com. 60 IN SRV 0 0 5667 lb.ati.com.lb.ati.com points to a local ip address in my hosts file.
@JigarJoshi, you can just call serverCall.sendMessage(). I think there are some caching interceptors floating around. This is client-side, and complicated, but still may be helpful. https://github.com/grpc/grpc-java/blob/master/examples/android/clientcache/app/src/main/java/io/grpc/clientcacheexample/SafeMethodCachingInterceptor.java
You can implement a global service via HandlerRegistry and pass it to serverBuilder.fallbackHandlerRegistry()
