-
Notifications
You must be signed in to change notification settings - Fork 178
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
java.io.IOException: The Application Default Credentials are not available in DataFlow pipeline #475
Comments
Is that error happening on your local machine or in the Dataflow environment? |
Executed at on DataFlow. JobId is In the logs also this warnings appear very often - Unsure if that is related:
|
The RejectedExecutionException happens when a connection is closed and more requests are occurring. I'm not sure why that would happen. I'll see if I can reproduce that problem outside of Dataflow. |
Ok. As |
Were able to reproduce. It looks like it can be 'enforced'/appears more likely if the BT servers are rebalancing (or the amount of BT servers it changed which also causes a rebalancing). Thanks to very few workers this should be the whole stacktrace:
I see where the This was logged to STDOUT - can this be intercepted at a better place and logged via SLF4j/commons so that it appears as one log entry in Cloud logging? |
According to the log, you are missing default application default credentials. Is this still happening? We need to talk to DataFlow to figure out why this is happening. |
We never saw it happening again except for this specific day. |
I don't believe #504 is related to this issue. The problem here is at startup time. If you don't see this problem anymore, we should close this. |
My best guess is that this was an intermittent Dataflow system configuration issue relating to Google Cloud Credential setup. Let's close this for now, since there is nothing actionable for Cloud Bigtable. We'll reopen if this becomes a problem. We'll let the Dataflow team know about this. |
Create a simple File/GCS to BigTable Dataflow pipeline:
Job is executed on 10 files around 3GB (+- 1.5GB) with autoscaling which jumps to 100 and 328 workers within 4 minutes.
We saw many times this - and it looks like that the same amount of messages didn't made it to BigTable (more precise investigation required)
The text was updated successfully, but these errors were encountered: