[grpc-io] Re: How to stop client streams correctly?

Yes. If you want to abort the stream, just cancel it with TryCancel. On Sunday, July 12, 2020 at 11:19:16 PM UTC-7 afshi...@gmail.com wrote: > > Hi. > > Let's assume that I can written an asynchronous grpc client that send a > stream and gets a response. This situation uses `ClientAsyncWriter`

[grpc-io] Re: GRPC Seg faults during streaming

This seems like some kind of memory corruption. I suggest you run the program under some memory tools such as AddressSanitizer. On Wednesday, May 13, 2020 at 3:04:06 AM UTC-7 deepankar wrote: > I am using grpc for streaming audio streams. My grpc client runs fine most > of the time, but

[grpc-io] Re: Name grpc threads

It is not clear which thread you are talking about and your platform. gRPC core does set thread names if applicable. For example the sync server will set its threads with name "grpcpp_sync_server" in thread_manager.cc. For linux, the name is set to the pthread if GPR_LINUX_PTHREAD_NAME macro

[grpc-io] Re: Stucking inside gRPC ClientReaderWriter's Read() function

You can turn on some debug tracing with environment variable GRPC_TRACE, for example setting it to "http,secure_endpoint,tcp". For a complete list, refer to: https://github.com/grpc/grpc/blob/master/doc/environment_variables.md On Saturday, October 26, 2019 at 11:24:07 AM UTC-7

Re: [grpc-io] Re: c++ multiple stubs

Channels are relatively heavier weight and thus it could be a good idea to not create a lot of them (unless you hit some throughput bottleneck). Stubs are pretty cheap. Channels and stubs are all thread-safe. On Tue, Nov 12, 2019 at 1:22 PM wrote: > Sorry, i meant "Is it better to also create

[grpc-io] Re: grpc server shutdown

It is safe to call Shutdown more than once. On Tuesday, August 20, 2019 at 3:18:52 PM UTC-7, Jeff wrote: > > If shutdown is called on a server more than once, is the behavior defined? > Is it safe to do this or is there a chance of a crash? -- You received this message because you are

Re: [grpc-io] What is the class do the function "Handshaker::DoHandShake()"

That is like a C++ virtual call and there would be a method on each handshaker. For example, the one you found is one of them. Not every handshaker is registered/run for every channel. You can run your client with GRPC_TRACE=handshaker GRPC_VERBOSITY=DEBUG and check the output log to see what

[grpc-io] Re: Memory leak possibility in greeter_async_client2.cc?

grpc completion queue will return all the tags you give it. To avoid the case you talk about, you will need to make sure you drain the completion queue before destroying it. On Wednesday, July 24, 2019 at 6:43:47 PM UTC-7, zhju...@gmail.com wrote: > > Hi, > > I'm trying to write an async grpc

[grpc-io] Re: correct way to shut down a synchronous server

You can call a grpc::Server::Shutdown with a timeout, after the timeout the pending rpcs will be cancelled. In your rpc handler (if you have long-lived streaming or rpcs), you will need to check ServerContext::isCancelled to finish up the rpcs. The sync server will not finish cleanup before you

[grpc-io] Re: A question about CompletionQueue under the use of async c++ client.

Other than Cq, grpc creates some internal threads to handle some background work. I guess it is possible that what you observed is that some work is offloaded to the background executor. On Friday, August 2, 2019 at 11:57:42 PM UTC-7, Arthur Wang wrote: > > Hi all: > >I know that the

Re: [grpc-io] How does ServerAsyncResponseWriter::Finish() in a async c++ server detect the timeout event ?

(adding back grpc-io) The client sends the relative timeout value ( https://github.com/grpc/grpc/blob/master/doc/PROTOCOL-HTTP2.md) over the wire to the server. The server receives the value and thus knows the timeout. The server can then set a timer to fire after the timeout to notify the rpc is

[grpc-io] Re: is byte-buffer returned by SerializationTraits::Serialize() read-only?

I think you can consider then read-only. The buffer is usually not changed in-place because you do not know whether you are the only owner that holds the ref. On Wednesday, June 26, 2019 at 9:22:21 AM UTC-7, Mayank Narula wrote: > > Hi folks > > I am referring to this definition - > /// 1.

[grpc-io] Re: Can I get reason for Unable to parse request(13)

The error message means the server could not parse the proto message. I would suggest you look into 1. whether server uses the same proto file with the same definition of the message 2. what is in the "ipc" field for failure case. On Tuesday, June 18, 2019 at 5:51:40 AM UTC-7, 윤석영 wrote: > >

[grpc-io] Re: [C++] Java onError equivalent in C++

If the server returns an error status, the read/write calls on the client side will fail and the status can be obtained via Finish. I do not think there is another way to do it. On Monday, May 13, 2019 at 7:16:47 AM UTC-7, dixi@gmail.com wrote: > > > Hi, > I have gRPC server with

[grpc-io] Re: gRPC url for service includes protobuf package name - how to change that?

This is documented in https://github.com/grpc/grpc/blob/master/doc/PROTOCOL-HTTP2.md I do not think there is an API to change it. On Wednesday, May 1, 2019 at 2:45:22 PM UTC-7, david...@xnor.ai wrote: > > [Using C++ bindings.] > > I was surprised to learn the hard way today that the gRPC URL

Re: [grpc-io] How to force gRPC to use Arena for memory allocation

I guess you are asking about synchronous server API. Unfortunately, there is no arena or general allocator support in a sync server. There are a couple of ways to workaround it if you really want to use arena: 1. You can switch to use an async server, where your application would control the

[grpc-io] Re: SSL connection failure(status:12, empty err message returned)

is your client and server using the same proto? Can you collect the client side trace as well? From the server log it does not seem like using ssl is related. On Thursday, September 13, 2018 at 5:44:22 PM UTC-7, solomon lifshits wrote: > > Thank you for reply! > No, the error message is empty.

Re: [grpc-io] Seg Fault when calling grpc::ServerInterface::RegisteredAsyncRequest::IssueRequest

The first thing I would check is whether server_ == nullptr after BuildAndStart. On Mon, Sep 10, 2018 at 3:22 PM Alistair Lowe wrote: > Hi guys, > > I'm trying to implement my own gRPC 1.14.2 C++ async server loosely > following the hello world examples, however I receive a segfault when >

[grpc-io] Re: [C++] How to detect client disconnect on synchronous streaming server

It sounds like the problem is similar to #15889 as you mentioned. We should continue the discussion in the issue. On Wednesday, July 18, 2018 at 10:47:51 AM UTC-7, banshe...@googlemail.com wrote: > > In principle, yes. I also already compiled a minimal reproducing example. > However I found

[grpc-io] Re: [C++] Server Mocking for Unit Testing (mock stream)

You are right. Currently there is no support for mocking server side sync reader/writer objects.. and it can only be tested with a real client and grpc library. On Friday, July 6, 2018 at 5:23:36 AM UTC-7, Alex Shaver wrote: > > When I have a service that has a simple unary interaction, writing

[grpc-io] Re: Multiple services registered to different ports / C++

My understanding is that you register the service with a host string to match a client using the corresponding authority for example registerService("other_service.myapi.com", service) and the client to use client_context.set_authority("other_service.myapi.com"); and then makes the rpc. At

[grpc-io] Re: [core/c++] Mocking timer for unit tests involving gRPC async calls.

I am afraid there is no easy way to do this now. In some tests we explicitly override the definition of "now". This might be able to achieve what you want, but usually it is used for something lower level. See bdp_estimator_test.cc for an example. On Friday, March 16, 2018 at 12:04:31 AM

Re: [grpc-io] Re: Non-blocking single-threaded streaming C++ server

Maybe you are over simplifying the code but the snippet does not look correct to me. 1. Your Read call is async, meaning it is more like StartRead and tell me via tag when there is something. So it does not make sense to log the args on its return. 2. You request a MdtDialOut but you do not seem

Re: [grpc-io] Re: gRFC L16 on new API to make gRPC C++ library initialization explicit

Hi Arpit, This discussion is not really related to this gRFC now, as it does not seem to be related to grpc_init/shutdown issue. If I recall correctly, what you did is correct. Let's say you have a thread pool driving the ServerCompletionQueue's to handle events for the async rpc, and in the

Re: [grpc-io] Re: Guidance on implementing support for authenticated and secure client connections

I do not think we support that. Actually I think tokens are intentionally dropped if the channel is not secure to avoid leaking the tokens over the wire. On Wed, Feb 7, 2018 at 3:04 PM, wrote: > I was able to get this working by simply calling AddMetadata on the client

Re: [grpc-io] Re: is NewStub (grpc_channel_register_call) leaking memory? (C++)

Hi, We will need some more details to understand what is going on. But I am not aware of any leaking in stub and channel. You can either provide some sort of repro or some debug information to show that memory has been leaked. Thanks. On Tue, Jan 30, 2018 at 10:11 PM, wrote:

[grpc-io] Re: [C++] limiting plugin option on ServerBuilder

Maybe you can use something like ServerBuilderSyncPluginDisabler in async_end2end_test.cc to remove the plugin for those servers. On Wednesday, December 13, 2017 at 4:41:32 PM UTC-8, Arpit Baldeva wrote: > > Hi, > > Currently, when the ServerBuilder::BuildAndStart is called, it > unilaterally

Re: [grpc-io] grpc c++ stream blocked when connect to an unavailable server

You can set a deadline to your stream. Re-issuing the rpc is not really a problem because a failed read at the client means the rpc is done at the server or failed somehow anyway. Or you can configure client side channel keep alive by adding channel arguments: GRPC_ARG_KEEPALIVE_TIME_MS and

[grpc-io] Re: C++ Graceful server shutdown

Do you mean you created 9 threads or grpc created 9 threads? I assume you are working on an async grpc server? Are those threads your rpc handling threads? grpc creates some threads internally to do some work offload. On Tuesday, November 28, 2017 at 2:19:28 AM UTC-8, shikhach...@gmail.com

[grpc-io] Re: I meet strange promble when test my own HelloWorld program.

For the first question, try to use 0.0.0.0:8889 at https://github.com/Rhysol/GrpcDemo/blob/master/server/main.cpp#L8 For the second question, your server handles a single rpc at a time. All the other rpcs are waiting at the server to be picked up. Maybe that is the reason why you see memory

[grpc-io] Re: [c++] What are the use of the threads that grpc starts in async mode?

gRPC core will start some threads to do background work (executor thread) and handle timers (timer thread). On Tuesday, November 7, 2017 at 12:38:56 PM UTC-8, Maysam Mehraban wrote: > > > Hello, > > I am looking into gRPC in async mode to see if it is suitable for the > application that I am

[grpc-io] Re: [grpc-c++] gRPC polling for incoming packets from multiple sockets

Yes. A ServerCompletionQueue is a CompletionQueue and you can use it for client events as well. On Wednesday, November 1, 2017 at 2:01:25 AM UTC-7, Ista Ranjan Samanta wrote: > > Hi, > > Thanks a lot for providing the accomplish able approach towards it. Yes, I > am also using C++ gRPC. > > I

[grpc-io] Re: [grpc-c++] About the destruction order in bidistream (Context and AsyncReaderWriter)

If your stream is finished, either way should be fine. To be on the safe side, destroy the reader/writer objects first and Client/ServerContext afterwards. On Thursday, September 7, 2017 at 10:59:47 AM UTC-7, yihao yang wrote: > > Hi, all, > > I have a question about the destruction order: > 1.

[grpc-io] Re: SSL - Missing selected ALPN property. CPP

Maybe try defining TSI_OPENSSL_ALPN_SUPPORT=0 and see whether it can fall back to NPN? On Monday, July 24, 2017 at 2:23:58 PM UTC-7, micha...@nauto.com wrote: > > My go client is set up like this and works correctly so I believe the > server and ELB are set up correctly: > > creds :=

[grpc-io] Re: [C++] CompletionQueue::Next() return vs ok parameter value

Usually we have a loop calling Next and when we are sure there will be no more work adding to the completion queue we call completion queue's Shutdown method to shutdown the queue. It will in turn cause the Next to return false and it can be used to break out of the loop. The ok parameter is

[grpc-io] Re: [C++] Status of tracing / census / OpenTracing integration?

There are currently non-public API under src/cpp/common/ for channel filters. You can try that out. I think its current form is not ready for moving to public headers. On Tuesday, July 25, 2017 at 4:09:15 PM UTC-7, Tudor Bosman wrote: > > Thanks! > > How about interceptors? > > Thanks, >

[grpc-io] Re: Asynchronous RPC to a CompletionQueue that has been shutdown

It is not clear to me what you meant by post an async rpc request to a cq. However, the user of a cq needs to guarantee that no new event is added to the cq after a shutdown is called. As a result, you most likely need some of your own synchronization. Thanks. On Thursday, July 27, 2017 at

[grpc-io] Re: Initializing the gRPC.

Can you give an example of the workflow that triggers the problem? We try to hide the initialization behind the creation of some high level C++ objects so that users should not need to worry about explicit initialization. Thanks. On Sunday, June 25, 2017 at 10:45:54 PM UTC-7,

Re: [grpc-io] Asynchronous Basics question

stream is part of your API definition while sync/async is about implementation. You should decide whether to have "stream" according to whether you can have more than one reply messages. As you said, whether to use async or sync implementation is about your resource limitations. If you have a

Re: [grpc-io] Re: How do I run grpc/test ?

You can make fling_client fling_server and look at fling_test.c to see how to run them manually. On Tue, Jun 13, 2017 at 9:02 AM, Rajarshi Biswas wrote: > Hi Guys, > > So I figured I have to make this using bazel ? But I get the following > error now. > > bazel

[grpc-io] Re: [ruby] request and require client certificate but don't verify

On the client side, I do not think you will have a way to completely disable checking. The closest you can get is to use ChannelArguments::SetSslTargetNameOverride to set the proper name from the server side cert. On Tue, Jun 6, 2017 at 3:05 PM, wrote: > Sorry for the delay

Re: [grpc-io] Re: How to set TOS bits on grpc server side for a new grpc client connection request?

On Wed, Apr 26, 2017 at 9:25 AM, Eric Anderson <ej...@google.com> wrote: > I assumed this is Go, given lis.Accept(). > > On Tue, Apr 25, 2017 at 5:14 PM, 'Yang Gao' via grpc.io < > grpc-io@googlegroups.com> wrote: > >> What language are you using? >> >

[grpc-io] Re: How to set TOS bits on grpc server side for a new grpc client connection request?

What language are you using? On Tuesday, April 18, 2017 at 4:30:24 PM UTC-7, Steven Jiang wrote: > > grpc server is listening on a port and accepting connection request from > grpc clients. lis.Accept() is called inside grpc lib. If I'd like to set > TOS bits of the new accepted connection,

Re: [grpc-io] Re: Retry policy

Ilina You may want to test by setting channel_arg with key grpc.testing.fixed_reconnect_backoff_ms to a smaller value to see whether it works. Or maybe playing with the values in client_channel/subchannel.c. I do not think we have a arg for the min_backoff right now though. On Tue, Feb 28, 2017

[grpc-io] Re: Exception if grpc::GoogleDefaultCredentials fails and error handling

What kind of error are you seeing? Are there any error logs printed? On Tuesday, October 18, 2016 at 1:16:27 AM UTC-7, Eugene Abramov wrote: > > Hello, > > If the file application_default_credentials.json is missing the function > grpc::GoogleDefaultCredentials fails with a critical error. How

[grpc-io] Re: No support for zookeeper_resolver any more

Hi Min, The zookeeper resolver code was not properly tested and caused us problems occasionally, and thus it was removed. If you are using that and have a tested implementation, we would be happy to take pull requests :) Thanks. On Wed, Nov 2, 2016 at 5:24 PM, Min Yao

Re: [grpc-io] is NewStub (grpc_channel_register_call) leaking memory? (C++)

Hi, The stub can be shared by different calls, but the memory is freed in https://github.com/grpc/grpc/blob/master/src/core/lib/surface/channel.c#L311 You can try to define GRPC_STREAM_REFCOUNT_DEBUG to have more logs. On Tue, Sep 27, 2016 at 5:17 PM, wrote: > I am tracking

[grpc-io] Re: Issue understanding how to properly stream to Cloud Speech API in C++ with grpc

You are talking to port 443 and using Insecure credentials (clear text). Should you use ssl credentials at least? On Monday, September 5, 2016 at 12:03:25 AM UTC-7, balad...@gmail.com wrote: > > Hello everyone! > > I've written some code in attempt to stream mic audio input from a > separate

Re: [grpc-io] detect client disconnect C++

You may want to try ServerContext::AsyncNotifyWhenDone. On Wed, Aug 17, 2016 at 2:51 AM, Chaitanya Gangwar < chaitanyagang...@gmail.com> wrote: > Hi, > > I have Async streaming server implemented in C++. > I want to know is there any way to detect on server side when a client > disconnects so