-
Notifications
You must be signed in to change notification settings - Fork 602
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
jenkins spark tests failing with 404 #3591
Comments
@davidbernick I opened this issue to track the problem. |
@mwalker174 Is encountering the same problem in the wild. He's reporting that it goes away if you specify the environment variables HELLBENDER_TEST_PROJECT and HELLBENDER_JSON_SERVICE_ACCOUNT_KEY he's seeing the warning message:
which should only appear during tests, so something is strange. |
As far as I can tell, getting that error message means that BaseTest is being loaded at runtime, and running it's static initializer block which calls |
I think the message isn't coming from BaseTest, its coming from a static block in SparkContextFactory:
|
@cmnbroad @lbergelson Looks like |
@jean-philippe-martin Can you comment on this one? It looks like
|
I was just talking with @vruano. The error might come from improper permissions/roles being set up on the cluster by default. |
It's weird that it worked before though if roles aren't set up right. It seems like security issues shouldn't be solved by asking people to upgrade their client software so that it can deny them permission. |
It seems plausible to me, though, that the Google auth library may have been patched to perform checks that it wasn't performing previously. Maybe our project permissions have always been mis-configured :) |
Sorry guys I have no special insight on this. Do you have a command line so I can try to reproduce locally? |
So this seems to only happen when trying to access a bucket from a job on dataproc. For example, the following throws the error: ./gatk-launch PathSeqFilterSpark -I gs://bucket/in.bam -O gs://bucket/out.bam -- --sparkRunner GCS --cluster my-cluster but the following does not: This happens even if I launch the cluster "gcloud dataproc clusters create ... --scope cloud-platform", which is supposed to grant full storage permissions. I believe this is equivalent to checking the "Allow API access to all Google Cloud Services" box if you launch a cluster through the web console. Also explicitly adding the service account as a "storage legacy bucket owner" does not seem to help. |
OK so just following along; the problem appears related to the Google Cloud Storage Connector and its configuration. When running on Cloud we need to ask for the The command line argument is Perhaps the code is trying to do the non-cloud setup and that's what's making it not work on cloud? |
this may be related to #3491, although that one predates this by quite a bit |
Apparently related, just running IndexFeatureFile on my machine results in several stack traces:
and
I was able to fix the issue by setting the environment variable |
@mwalker174 @lbergelson @jean-philippe-martin I was able to fix the 404 error by building a custom version of the lastest master of
So it's likely the |
I've updated googleapis/google-cloud-java#2453 with this result -- we'll see what they say. |
Fixed in |
…f of our custom fork The google-cloud-java maintainers have merged a fix for the longstanding issue googleapis/google-cloud-java#2453 that prevented us from running on a modern version of the library, and forced us to run off of a fork. This PR updates us to the latest release, which incorporates the fix. Resolves #3591 Resolves #3500 Resolves #4986
…f of our custom fork (#5135) The google-cloud-java maintainers have merged a fix for the longstanding issue googleapis/google-cloud-java#2453 that prevented us from running on a modern version of the library, and forced us to run off of a fork. This PR updates us to the latest release, which incorporates the fix. Resolves #3591 Resolves #3500 Resolves #4986
The jenkins spark tests are failing with the following error:
This seems to have been introduced in #3576
The text was updated successfully, but these errors were encountered: