If Slack client closes, exit the hubot process

[eriklindebratt]

Since Slack client seems to have troubles reconnecting occasionally (e.g. #203), this seems like the only way to solve it right now. I'm running Hubot using supervisor and in the case of Slack client dropping its connection it seems to me that having the Hubot process exit and get restarted from e.g. supervisor is the way to go.

[farski]

Would this be better as an optional behavior change?

[knpwrs]

I'd like to see this merged but I agree with @farski that it should be behind an option.

[eriklindebratt]

@farski @KenPowers Yep, I agree. Would you prefer something like an env var (e.g. HUBOT_SLACK_EXIT_ON_DISCONNECT=true bin/hubot --adapter slack)?

[knpwrs]

Either that or if hubot could support subargs then we could have:

bin/hubot --adapter slack [ --exit-on-disconnect ]

But I think environment variables are more likely.

[christopherRiddersater]

Sorry for jumping in like this.
Behind an option would be nice and as @KenPowers points out, a subarg would be nice but either would be a great improvement.

[eriklindebratt]

@farski @KenPowers @christopherRiddersater I added a check for an environment variable HUBOT_SLACK_EXIT_ON_DISCONNECT and that is used as a flag in slack.coffee on whether the Hubot process should exit when the Slack client disconnects. Thoughts?

[farski]

@eriklindebratt Looks reasonable to me. Only thing is exit_process_on_disconnect should probably be exitProcessOnDisconnect by convention.

[eriklindebratt]

@farski Ah missed that one, thanks.

[christopherRiddersater]

Looks good to me.

[spoon99]

+1

[bradam12]

+1 This would help my monitoring for sure. My hubot tends to not connect back, so it sits there not doing anything.

[evansolomon]

Is this issue pretty much the opposite of #127?

[bradam12]

Seems so, but is configurable on what behavior you want. If you want it to try reconnecting, leave it as is. If you want it to exit so supervisor/script/equivalent can detect it died and spin Hubot back up, go for it and turn the option on.

[knpwrs]

The behavior I was observing is that my hubot would never reconnect.

[bradam12]

Same. Mine never tries reconnecting (seemingly), so I'd rather let it exit and respin it back up.

[ivarrian]

Same as @KenPowers . The process is up but the last log I see is INFO Slack client closed, waiting for reconnect , but it doesn't ever reconnect. I have setup foreverjs to restart the process when it dies, but it actually sticks around doing nothing.

[knpwrs]

That said, the preferable solution for the long term would be to figure out why hubot isn't reconnecting.

[bradam12]

Agreed.

[evansolomon]

Seems reasonable to me

[christopherRiddersater]

Yes, I agree about the long term solution and I think that the short time solution may prove useful for other scenarios in the future as well.

[mjsz]

This looks reasonable as a stopgap, thanks all. I will plan to merge and do a patch release with this included later today.

[eriklindebratt]

Sweet @mjsz, thanks!

[christopherRiddersater]

@mjsz Great! Many thanks!

[MrJoy]

Any notion of when a new release that includes this will be pushed to NPM?