Creating Channel stucks on the unstable network

[KimTaehee]

Version: 0.14.2
Severity: Critical
Symptom: ApnsChannelPool cannot acquire Channel forever
Properties: concurrentConnections=1, eventLoopGroup=NioEventLoopGroup(1 thread)
Region: China(once every 2 days), Europe(once every 7 days)

Hi, Jon. I'm considering applying Pushy in the global environment. And I really appreciate your efforts.
Sadly I often faced a problem above with this so I'm trying to find the cause.

I believe these are the correct steps to reproduce in debug mode:

1. Break at the last line of io.netty.channel.nio.AbstractNioChannel#connect()
2. Down the network (ifconfig down or whatever)
3. Resume application
4. After few(<30?) seconds later, Netty calls ApnsClientHandler#channelInactive()
5. Resume the network
6. Then ApnsChannelPool#pendingCreateChannelFutures have 1 createChannelFuture that has not been done
7. ApnsChannelPool#acquireWithinEventExecutor() can no longer create the proper Channel

Of course, we can reproduce this symptom in normal mode.

I think we must finish the createChannelFuture so I suggest inserting lines on ApnsClientHandler#channelInactive() to resolve pending createChannelFuture.

I look forward to hearing from you. Thanks.

[janzar]

Can confirm that it is repeatable. just small typo:

1. at the last line of io.netty.channel.nio.AbstractNioChannel#connect()

[KimTaehee]

Can confirm that it is repeatable. just small typo:

1. at the last line of io.netty.channel.nio.AbstractNioChannel#connect()

Thanks @janzar , I updated.

[jchambers]

Thank you for the fix in #854! I've merged that, and hope that will resolve this issue.