0

When I try to deploy Datalab to one of my existing projects, I am running into "Giving up due to transient errors" in the startupscript (second to last line in the log output tail below). All attempts log themselves in the automatically created PROJECT-datalab/logs storage bucket, and the very first attempt has some additional content at the start (one-time bootstrapping I believe), but all fail the same way.

I can also see a datalab module dashboard under the cloud console. If I return to https://datalab.cloud.google.com/ and select the project the only possible option remains "Deploy Datalab"; "Start Datalab" and "Manage Datalab" are greyed out.

Here is the tail of the log output (identical aside from timestamps for all the log files I checked):

Nov 10 01:43:43 datalab-deploy-main-20151110-01-41-53 startupscript: You are about to deploy the following modules:
Nov 10 01:43:43 datalab-deploy-main-20151110-01-41-53 startupscript:  -    mezuri-dev/datalab/main  From: [/datalab/app.yaml]
Nov 10 01:43:43 datalab-deploy-main-20151110-01-41-53 startupscript: Verifying that Managed VMs are enabled and ready.
Nov 10 01:43:47 datalab-deploy-main-20151110-01-41-53 startupscript: If this is your first deployment, this may take a while...#015If this is your first deployment, this may take a while.../#015If this is your first deployment, this may take a while...-#015If this is your first deployment, this may take a while...\#015If this is your first deployment, this may take a while...|#015If this is your first deployment, this may take a while...done.
Nov 10 01:43:47 datalab-deploy-main-20151110-01-41-53 startupscript: 
Nov 10 01:43:47 datalab-deploy-main-20151110-01-41-53 startupscript: Building and pushing image for module [datalab]
Nov 10 01:44:13 datalab-deploy-main-20151110-01-41-53 startupscript: Updating module [datalab]...#015Updating module [datalab].../#015Updating module [datalab]...-#015Updating module [datalab]...\#015Updating module [datalab]...|#015Updating module [datalab].../#015Updating module [datalab]...-#015Updating module [datalab]...\#015Updating module [datalab]...|#015Updating module [datalab].../#015Updating module [datalab]...-#015Updating module [datalab]...\#015Updating module [datalab]...|#015Updating module [datalab].../#015Updating module [datalab]...-#015Updating module [datalab]...\#015Updating module [datalab]...|#015Updating module [datalab].../#015Updating module [datalab]...-#015Updating module [datalab]...\#015Updating module [datalab]...|#015Updating module [datalab].../#015Updating module [datalab]...-#015Updating module [datalab]...\#015Updating module [datalab]...|#015Updating module             [datalab]...done.
Nov 10 01:44:13 datalab-deploy-main-20151110-01-41-53 startupscript: ERROR: (gcloud.preview.app.deploy) Giving up due to transient errors. Please try again. Version: datalab:main.388458272080065440
Nov 10 01:44:14 datalab-deploy-main-20151110-01-41-53 startupscript: Step deploy datalab module failed.

Any idea what the problem may be? This "transient" error has occurred in each of 10 attempts spread out over the last 4 hours. Is this perhaps one of those existing projects that don't currently work without manual intervention?

Incidentally if I follow the log link from the "Deploy Datalab" page it goes through a cycle of:

not found

then changes to

Forbidden

and finally ends at

{"error":{"errors":[{"domain":"global","reason":"lockedDomainExpired","message":"Locked Domain Expired"}],"code":401,"message":"Locked Domain Expired"}}

I think that actually might be a multiple-Google-accounts issue and unrelated, as the complete log files (without that 401 message) do appear in the storage bucket, but mention it just in case.

mpod
  • 1

1 Answers1

0

Sorry to hear it doesn't work in your existing projects. If possible, can you create a new project and see if deployment works there?

For your existing project, I am not sure if it is really a transient issue. Since it already created a datalab module in your project, we might be able to get the console output from the managed VM instance to see what might have failed. To do so, remove your existing datalab module, and then reproduce the issue one more time. Then, go to Developer Console, Compute Engine, VM Instances. find the instance that corresponding to datalab (something like gae-datalab-main-****, if it exists), click on it and browse to the end of the page, click Serial console output. See if there is anything interesting there?

Bradley Jiang
  • 424
  • 2
  • 1
  • Thanks Bradley. In trying to deploy it to a brand new project, I ran into the same 403 error as described here: http://stackoverflow.com/questions/33583437/datalab-cloud-deployment-403-error , but by following your answer there I was able to successfully deploy it. So it's working on a fresh project. Could you expand on how to remove the existing datalab module? I see it under the App Engine dashboard but don't see any removal options, and there are no VM instances. Also in case it matters: the existing project has a numeric ID (so not project-name-NNNN-compute@... user) and is in a CDF EAP. – mpod Nov 11 '15 at 05:53