This message was deleted.
# general
s
This message was deleted.
s
run a simple batch ingestion and see if you get segments in the bucket
initially, streaming ingestion will not publish segments, it will publish at handoff which occurs normally at the end of the task duration (or at intermediateHandoffPeriod).
a
I normally test with the wikipedia sample from the web console loader.
v
👍 will give a try, thanks!
Auth failure now. And I see
WebIdentityTokenCredentialsProvider
is acceptable in the provider chain here. AWS_ROLE_ARN, AWS_WEB_IDENTITY_TOKEN_FILE are available in the env.
Copy code
2023-03-28T17:55:54,309 WARN [[index_parallel_wikipedia_nhgflfaj_2023-03-28T17:55:31.446Z]-appenderator-merge] org.apache.druid.java.util.common.RetryUtils - Retrying (1 of 4) in 1,008ms.
java.io.IOException: com.amazonaws.services.s3.model.AmazonS3Exception: Access Denied (Service: Amazon S3; Status Code: 403; Error Code: AccessDenied; Request ID: NETB5ZBG7E7FH9Z7; S3 Extended Request ID: iZluZGHneBhidh5UvQYDvTyJL6iznH89s3WPtoZB2BakmnLm5QblX1SuIKEIrjX1f7md18/sTDcyC7Dx3cKgEw==; Proxy: null), S3 Extended Request ID: iZluZGHneBhidh5UvQYDvTyJL6iznH89s3WPtoZB2BakmnLm5QblX1SuIKEIrjX1f7md18/sTDcyC7Dx3cKgEw==
	at org.apache.druid.storage.s3.S3DataSegmentPusher.pushToPath(S3DataSegmentPusher.java:108) ~[?:?]
	at org.apache.druid.storage.s3.S3DataSegmentPusher.push(S3DataSegmentPusher.java:82) ~[?:?]
	at org.apache.druid.segment.realtime.appenderator.AppenderatorImpl.lambda$mergeAndPush$4(AppenderatorImpl.java:956) ~[druid-server-25.0.0.jar:25.0.0]
a
If you describe the pod this is running on do you see the AWS related environment variables set? Did you use the Druid Helm chart to install or some other method? Did you specify the IRSA annotation on the service account for all service types?
v
yes, pod does have env vars. I am using operator to create this cluster, yes, druid helm charts.. serviceaccount has an annotation
Copy code
Annotations:         <http://eks.amazonaws.com/role-arn|eks.amazonaws.com/role-arn>: arn:aws:iam::<<myaccount>>:role/kubernetes_services/<<myrole>>
a
Does the role have access to read from the S3 bucket on both the bucket and the role?
Well read and write most likely is required.
v
yes, I am able to read/write from a test pod, but not druid
wondering if there is a specific rbac needed for the pods
f
have you set up the druid service account in druid-operator?eg. serviceAccount: druid-application-sa, you need to add S3 permissions to the service account.
v
never mind, just enabled trace logging on
com.amazonaws
and noticed that the secret file is not being picked up. This is our k8s setup issue where token file secret is not mounted to druid pods.
Copy code
2023-03-28T21:57:58,829 DEBUG [[index_parallel_wikipedia_acdefclb_2023-03-28T21:57:40.714Z]-appenderator-merge] com.amazonaws.auth.AWSCredentialsProviderChain - Unable to load credentials from WebIdentityTokenCredentialsProvider: Unable to locate specified web identity token file: /var/run/secrets/eks.amazonaws.com/serviceaccount/token