Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

test: grpc_client_test and fix grpc_client StopForError() hang bug #155

Merged
merged 1 commit into from
Dec 29, 2020
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion abci/client/grpc_client.go
Original file line number Diff line number Diff line change
Expand Up @@ -88,11 +88,11 @@ func (cli *grpcClient) OnStop() {
}

func (cli *grpcClient) StopForError(err error) {
cli.mtx.Lock()
if !cli.IsRunning() {
return
}

cli.mtx.Lock()
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think cli.IsRunning() and cli.Stop() should be protected by the mutex. Otherwise, the error might be overwritten by a concurrent thread.

cli.mtx.Lock()
defer cli.mtx.Unlock()
if !cli.IsRunning() {
  return
}
cli.Stop()
...

Copy link
Author

@jinsan-line jinsan-line Dec 28, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I had same thoughts when I met this issue first. But, I choose current implementation because,

  • I refer the reference implementation from socket_client. (I've updated it in PR description)
  • I wonder why the IsRunning() and Stop() are not protected by the mutex but I realize that they are already atomic.

Reference:
https://github.com/line/tendermint/blob/a01efd28c7257d3b056d1fcf263d39ccea3c1273/libs/service/service.go#L202-L206

https://github.com/line/tendermint/blob/a01efd28c7257d3b056d1fcf263d39ccea3c1273/libs/service/service.go#L159-L175

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK, I understand.
There is a trivial problem that Stopping abci.grpcClient for error: ... log message might be written twice by two separate concurrent threads, but it's OK.

if cli.err == nil {
cli.err = err
}
Expand Down
93 changes: 93 additions & 0 deletions abci/client/grpc_client_test.go
Original file line number Diff line number Diff line change
@@ -0,0 +1,93 @@
package abcicli_test

import (
"errors"
"fmt"
"testing"
"time"

"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"

abcicli "github.com/tendermint/tendermint/abci/client"
"github.com/tendermint/tendermint/abci/server"
"github.com/tendermint/tendermint/abci/types"
tmrand "github.com/tendermint/tendermint/libs/rand"
"github.com/tendermint/tendermint/libs/service"
)

type errorStopper interface {
StopForError(error)
}

func TestSocketClientStopForErrorDeadlock(t *testing.T) {
c := abcicli.NewGRPCClient(":80", false).(errorStopper)
err := errors.New("foo-tendermint")

// See Issue https://github.com/tendermint/abci/issues/114
doneChan := make(chan bool)
go func() {
defer close(doneChan)
c.StopForError(err)
c.StopForError(err)
}()

select {
case <-doneChan:
case <-time.After(time.Second * 4):
t.Fatalf("Test took too long, potential deadlock still exists")
}
}

func TestProperSyncCalls(t *testing.T) {
app := slowApp{}

s, c := setupClientServer(t, app)
defer s.Stop()
defer c.Stop()

resp := make(chan error, 1)
go func() {
// This is BeginBlockSync unrolled....
reqres := c.BeginBlockAsync(types.RequestBeginBlock{})
c.FlushSync()
res := reqres.Response.GetBeginBlock()
require.NotNil(t, res)
resp <- c.Error()
}()

select {
case <-time.After(time.Second):
require.Fail(t, "No response arrived")
case err, ok := <-resp:
require.True(t, ok, "Must not close channel")
assert.NoError(t, err, "This should return success")
}
}

func setupClientServer(t *testing.T, app types.Application) (
service.Service, abcicli.Client) {
// some port between 20k and 30k
port := 20000 + tmrand.Int32()%10000
addr := fmt.Sprintf("localhost:%d", port)

s, err := server.NewServer(addr, "grpc", app)
require.NoError(t, err)
err = s.Start()
require.NoError(t, err)

c := abcicli.NewGRPCClient(addr, true)
err = c.Start()
require.NoError(t, err)

return s, c
}

type slowApp struct {
types.BaseApplication
}

func (slowApp) BeginBlock(req types.RequestBeginBlock) types.ResponseBeginBlock {
time.Sleep(200 * time.Millisecond)
return types.ResponseBeginBlock{}
}