internal/poll.runtime_pollWait panic when high concurrent write to redis

[JobberRT]

Hi guys, i've a program which has around 500~1000 concurrency write/read, and i got this panic log

goroutine 1526 [runnable]:
internal/poll.runtime_pollWait(0x9ba30c0, 0x72)
        /Users/jobber/sdk/go1.17.3/src/runtime/netpoll.go:234 +0x89
internal/poll.(*pollDesc).wait(0xc00695e200, 0xc005efb000, 0x0)
        /Users/jobber/sdk/go1.17.3/src/internal/poll/fd_poll_runtime.go:84 +0x32
internal/poll.(*pollDesc).waitRead(...)
        /Users/jobber/sdk/go1.17.3/src/internal/poll/fd_poll_runtime.go:89
internal/poll.(*FD).Read(0xc00695e200, {0xc005efb000, 0x1000, 0x1000})
        /Users/jobber/sdk/go1.17.3/src/internal/poll/fd_unix.go:167 +0x25a
net.(*netFD).Read(0xc00695e200, {0xc005efb000, 0xc00004a800, 0xc0007bb318})
        /Users/jobber/sdk/go1.17.3/src/net/fd_posix.go:56 +0x29
net.(*conn).Read(0xc0002f50e8, {0xc005efb000, 0xc006963000, 0x59})
        /Users/jobber/sdk/go1.17.3/src/net/net.go:183 +0x45
bufio.(*Reader).fill(0xc005ef80c0)
        /Users/jobber/sdk/go1.17.3/src/bufio/bufio.go:101 +0x103
bufio.(*Reader).ReadSlice(0xc005ef80c0, 0x5a)
        /Users/jobber/sdk/go1.17.3/src/bufio/bufio.go:360 +0x2f

// Some code in my program

And here's how i use go-redis

// Init the rdb
rdb = redis.NewClient(&redis.Options{
		Addr:         viper.GetString("redis.address"),
		Password:     viper.GetString("redis.password"),
		DB:           viper.GetInt("redis.db"),
		PoolSize:     1000,
		ReadTimeout:  30 * time.Second,
		WriteTimeout: 30 * time.Second,
	})

// Using the rdb
notExist, err := rdb.SetNX(ctx, SomeString, true, time.Minute).Result()
	if err != nil {
		logrus.Error(err)
		return 0
	}
	if !notExist {
		logrus.Debugf("%s exist, skip", p.Host)
		return 0
	}

Did i do something wronog? What should i do to avoid this panic? I've tried increase the pool size but it doesn't help.

[vmihailenco]

You haven't shown the panic message.

[JobberRT]

Ohhhh, sorry! I forget to copy the panic message! Found that IDEA GoLand's output buffer size is small at default and flush the panic message! After increase the buffer size, i can get the panic message and i've already sovle this, turns out that this should be my own code's problem, thanks!

[jonliubj]

Hi @JobberRT , I am having the same issue with high concurrency (1000 goroutines). Can you briefly share what is your code's problem that causes this issue? Many thanks!!

[JobberRT]

Sorry, it's a long time and I can't find my history. But I'm pretty sure it's my problem rather than the library.