Use of closed tcp connection upon message send about slack HOT 6 CLOSED

Same issue here, when sending a message through RTM with the the following code:

rtm.SendMessage(rtm.NewOutgoingMessage(mytext, IDchannel))

The RTM event I receive:

event.type: "outgoing_error"
event.data: write tcp 10.0.0.212:57025->54.172.82.211:443: i/o timeout

EDIT:
There is not the issue when using this code:

api.PostMessage(IDchannel, mytext,slack.PostMessageParameters{AsUser: true})

from slack.

Worked fine before, but started a few days ago. This code:

rtm.SendMessage(rtm.NewTypingMessage(event.Channel))

now results in i/o timeout and disconnect:

OutgoingError: write tcp 172.*.*.*:41407->54.225.22.47:443: i/o timeout
Disconnected: intentional false
IncomingError: read tcp 172.*.*.*:41407->54.225.22.47:443: use of closed network connection

Looks like RTM hits some new rate limits, and Slack closes connection. I had to turn off typing notifications. api.PostMessage works fine.

from slack.

So I'm using this wrapper for the bot, which uses rtm.SendMessage: https://github.com/BeepBoopHQ/go-slackbot/blob/master/bot.go#L116, this one times out regularly. @ignat you're saying if I replace that with api.PostMessage it would work?

That's cool, I'm just curious what we can do for this lib to fix RTM as well.

from slack.

I think this is related with nlopes@b7940f2.

I'm reverting until I have some time to look into these issues.

from slack.

Reverted branch works well indeed.

from slack.

I can confirm that this commit caused the problem: nlopes@8955910

Also, it seems that we're attempting to set the write deadline on the PING only where in actuality when using SetWriteDeadline it affects any writes thereafter on the current connection.

So there are two issues:

• the one second deadline is too short
• SetWriteDeadline should probably be moved to ManageConnection()

I just submitted this PR #102 to resolves these issues.

from slack.