Deployments keep on failing, localhost:8080 issues

[Kaan Soral]

This used to happen every once in a while, but lately it happens 99% of the time, the browser auth process just fails to complete, it halts, redirects to http://localhost:8080/ that doesn't respond

Any ideas?

[Nick (Cloud Platform Support)]

Hey Kaan,

This doesn't appear to contain enough information to constitute an effective start for any investigation from our side. If you believe this is an issue related to the platform code, you could file an issue in the Public Issue Tracker, probably with a HAR recording of the browser activity during the interaction, and any command line output, along with relevant info such as the version of the SDK. Be sure to attach the label "Restrict-View-EditIssue" to ensure that your HAR information is not exposed publicly. With that information, we'd be in a position to look further.

Cheers,

Nick
Cloud Platform Community Support 

[Kaan Soral]

How does the deployment auth work?

Why does it fail at times and work flawlessly at other times?

Why does it redirect to localhost:8080?

You could at least give some hints

[Kaan Soral]

This started being a regular issue for me, now the auth's redirect to :8090 - I'm using the port 8090 myself, the SDK uses the 8080

The initial deployment usually works, but for re-deployments, I'm usually hit by this issue :(

[Nick (Cloud Platform Support)]

Hey Kaan,

I think there's an unstated assumption here which is creating a bit of difficulty in getting to a consensus on the next steps. We don't expect that browser auth for gcloud should be redirecting to localhost:8080. You should be given a URL like the following:

https://accounts.google.com/o/oauth2/auth?redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&prompt=select_account&response_type=code&client_id=12345678.apps.googleusercontent.com&scope=https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fuserinfo.email+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fcloud-platform+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fappengine.admin+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fcompute&access_type=offline

In order to determine what's going on, since it's not quite clear really what's even going on, let alone why it might be going on, we'll need some precise technical details that can be given in a Public Issue Tracker report as mentioned above. The precise code / commands relevant along with HAR recordings of the browser activity, or a tcp dump of the browser's activity during that time, should be sufficient.

Let me know when you've made a Public Issue Tracker report so I can take a look at it right away, although of course we do aim to respond to threads as soon as possible and monitor the forums daily.

Cheers,

Nick

Cloud Platform Community Support

[Kaan Soral]

Hi

Thanks for the reply

This is the url I get: https://accounts.google.com/o/oauth2/auth?scope=https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fappengine.admin+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fcloud-platform+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fuserinfo.email&redirect_uri=http%3A%2F%2Flocalhost%3A8080%2F&response_type=code&client_id=XXX.apps.googleusercontent.com&access_type=offline

I use appcfg.py to deploy, I think it creates a temporary webserver for deployments, which get stuck, this is my theory

Some findings:

1) Shutting down the dev_appserver which natively uses 8080 port doesn't solve the issue

2) The issue usually happens on re-deployments, when it happens, it usually last 10 to 30 minutes

I usually deploy in heat, sometimes a new deployment causes an issue, you need to re-deploy fast, this issue steals a lot of precious time that would otherwise be productive

I was using the 1.9.36 SDK, switched to 1.9.40, I don't know whether it fixes the issue

In any case, I just tried `--noauth_local_webserver` - should've tried it earlier, so my wasted time is on me

I'm speculating there is a temporary webserver that just serves to take that auth code from the browser and feed it into appcfg, that webserver glitches a lot (at least it did on 1.9.36, I didn't get to test whether things are same on 1.9.40)

[Nick (Cloud Platform Support)]

Hey Kaan,

That does indeed sound correct! If you could manage to reproduce this behaviour any time in future, or if anybody finds this thread who is experiencing a similar issue, it would be very helpful to get the issue logged with as much detail in the Public Issue Tracker for App Engine, that way we could work on fixing whatever seems to be the issue. Glad to hear you got a workaround with --noauth_local_webserver, though. 

Cheers

Nick
Cloud Platform Community Support