Slack error messages

changed milestone to %9.1 in gitlab-foss

The real fix is to prioritise performance improvements. Other than that, I am not aware of any changes introduced in 9.1

@zj what's the timeout?

We see the same thing. Since 9.1 was deployed we get the timeout on issue create. We get the timeout on pretty much every slash command. issue new does create the issue (along with displaying the error message), and "help", "issue show" just return the timeout

Before 9.1 was deployed we would occasionally see this behavior, but generally did not. Now it's a no fail timeout insofar as I can see. If nothing about 9.1 changed, maybe there was an infrastructure change that might be related? Didn't you have some sort of unintended Redis upgrade happen during that deploy... pure speculation and I know nothing of your architecture, but something is markedly different once the 9.1 release was deployed. Wish I could help more.

I'm just trying out GitLab's Slack Slash commands for the very first time, and we haven't been able to get this working at all over the last few days; we're consistently seeing 'Timeout was reached'.

I think that this is probably a performance issue, and how GitLab just isn't responding to Slack quickly enough. This is evidenced by response rates ranging from 4000ms - 6000ms (Slack's timeout is 3000ms).

$ time http POST https://gitlab.com/api/v3/projects/SomeFakeProject/services/slack_slash_commands/trigger token=SomeFakeToken
HTTP/1.1 200 OK
Cache-Control: max-age=0, private, must-revalidate
Content-Length: 112
Content-Type: application/json
Date: Tue, 02 May 2017 22:54:16 GMT
Server: nginx
Strict-Transport-Security: max-age=31536000
Vary: Origin
X-Request-Id: dd7bc0e6-9e0a-467e-b00b-c96510fa223a
X-Runtime: 0.218490

{
    "response_type": "ephemeral",
    "status": 200,
    "text": ":sweat_smile: Couldn't identify you, nor can I autorize you!"
}


real	0m6.108s
user	0m0.246s
sys	0m0.045s

(on a side note, there's a typo (e.g., authorize), and I'm also surprised it's a 200 response code and not something else, but meh, either way it doesn't work)

NOTE: I'm assuming that putting all of Slack's data in the request isn't leading to the delay.

If performance is indeed the problem, would a short-term fix be to use the Delayed Response feature of Slack's slash commands? Ref: https://api.slack.com/slash-commands

@zj Out of curiosity, are you seeing this problem in 9.1 (specifically public GitLab.com)?

@smcgivern The timeout is 2 seconds.

I have not yet debugged this, and sadly until next week at least I don't have time for this. I do feel there is a lot of regression in terms of performance in this feature.