🐛 Bug Report

The TestTimeout test is sometimes failing randomly.

To Reproduce

$ go test -run TestTimeout -count 100 ./examples/internal/integration

Expected behavior

Test not to be flaky

Actual Behavior

The tests fails with an incorrect status code error:

integration_test.go:1395: resp.StatusCode = 500; want 504

See https://app.circle

Before you submit this issue, you has been search all existed issues and search the documentation

• [] I've been search all existed issues
• [] I've been read all documentation

Describe your question

How should I use alicloud MNS?

In our codebase, we push RequestContext and immediately run some code with try-with-resources.
For example:

try (SafeCloseable ignored = ctx.push()) {
    logger.trace(decorate(msg));
}

If RequestContext provides run(Runnable) or call(Callable) we can reduce boilerplate code and simplify it.

ctx.run(() -> logger.trace(decorate(msg));

This is inspired by g

virtual host options

We should also hook up a grpc-go client to test the interop with the tonic server.

Currently, in https://github.com/hyperium/tonic/blob/master/interop/test.sh we only test tonic client -> grpc-go server and tonic client to tonic server. We should add an additional client test that will go from a grpc-go client to tonic server.

https://github.com/douyu/jupiter/blob/f8efede954c15a815a238feb64e2fc726600be9a/pkg/xlog/buffer_syncer.go#L62-L88

time.NewTicker 在使用过程中，结束的时候一定要 stop。

• 官方 Example 如下：

ticker := time.NewTicker(time.Second) 
defer ticker.Stop() 
done := make(chan bool) 
go func() {     
    time.Sleep(10 * time.Second)     
    done <- true 
}() 
for {     
    select {
     case <-done:
      

In integration tests today we don't test properly that reset API wipes the node, we just test for reboot.

We could have used filesystem UUID change to verify that node got wiped and re-installed. (thanks @Ulexus)