georgew5656 opened a new pull request, #15287:
URL: https://github.com/apache/druid/pull/15287

   ### Description
   Druid does not support various forms of auth that Azure Storage Accounts has 
added support for, like Service Client auth, Workload Identity Auth (e.g. for 
AKS) or managed identity Auth. Adding the AzureDefaultCredential chain to the 
azure clients druid uses allows support for configuring these things. See: 
https://learn.microsoft.com/en-us/java/api/overview/azure/identity-readme?view=azure-java-stable
   
   In order to support this new identity library, I had to upgrade the druid 
azure client libraries for blob storage (they were very old anyways).
   
   #### Fixed the bug ...
   #### Renamed the class ...
   #### Added a forbidden-apis entry ...
   - Many of the changes from the client upgrade were pretty straightforward (I 
used this .NET docs a lot: 
https://github.com/Azure/azure-sdk-for-net/blob/main/sdk/storage/Azure.Storage.Blobs/AzureStorageNetMigrationV12.md#class-conversion-reference)
   - Most of these changes are in AzureStorage
   Less straightforward changes
   - The client supports retrying requests by default, so I removed instances 
of RetryUtils in favor of using the client directly. There are several places 
where druid uses a different retry configuration for different azure 
operations, so I had to add AzureClientFactory to support creating additional 
clients on the fly (because the library doesn't let you override the retry 
policy on existing clients). I also updated AzureStorage to reuse these 
per-container overrided clients, the new client is stateless so it is 
technically fine to just recreate a client each time but i figured it would be 
better to reuse them as a performance optimization. A couple of the tests were 
removed because they were testing this retry functionality
   - I had to change the AzureCloudBlobIterator because the new client supports 
pagination by default (it only fetches additional pages if you request 
additional items)
   #### Release note
   - Upgrade druid azure client libraries
   - Support AzureDefaultCredential for druid auth to azure services
   - I removed some of the mocks from the tests and added a test implementation 
(TestPagedResponse). This let me remove the ListBlobItemHolder class that was 
mostly there for testing.
   
   ##### Key changed/added classes in this PR
   * AzureClientFactory
   * AzureAccountConfig
   * AzureCloudBlobIterator
   * AzureStorage
   * AzureStorageDruidModule
   <hr>
   
   <!-- Check the items by putting "x" in the brackets for the done things. Not 
all of these items apply to every PR. Remove the items which are not done or 
not relevant to the PR. None of the items from the checklist below are strictly 
necessary, but it would be very helpful if you at least self-review the PR. -->
   
   This PR has:
   
   - [X] been self-reviewed.
      - [ ] using the [concurrency 
checklist](https://github.com/apache/druid/blob/master/dev/code-review/concurrency.md)
 (Remove this item if the PR doesn't have any relation to concurrency.)
   - [ ] added documentation for new or modified features or behaviors.
   - [ ] a release note entry in the PR description.
   - [ ] added Javadocs for most classes and all non-trivial methods. Linked 
related entities via Javadoc links.
   - [ ] added or updated version, license, or notice information in 
[licenses.yaml](https://github.com/apache/druid/blob/master/dev/license.md)
   - [ ] added comments explaining the "why" and the intent of the code 
wherever would not be obvious for an unfamiliar reader.
   - [X] added unit tests or modified existing tests to cover new code paths, 
ensuring the threshold for [code 
coverage](https://github.com/apache/druid/blob/master/dev/code-review/code-coverage.md)
 is met.
   - [ ] added integration tests.
   - [X] been tested in a test Druid cluster.
   
   I have tested the following scenarios for auth
   SAS token auth
   Key auth
   Workload identity auth (using a k8s serviceAccount and managed identity to 
mount credentials for druid running on a pod in AKS)
   Environment variable auth (using a App registration client id/secret)
   Managed identity auth (I technically have not tested this directly on a VM, 
but I am pretty sure this will work on a VM with a managed identity by setting 
the managedIdentityClientId property 
https://github.com/Azure/azure-sdk-for-java/wiki/Azure-Identity-Examples#authenticating-in-azure-with-managed-identity)
   Technically the AzureDefaultCredential supports more types of auth but I did 
not test any of them.
   
   I also tested the following migration scenarios
   Cluster running old azure extension using a sas token -> new azure extension 
using a sas token
   Sas token (new client) -> workload credential auth (new client)
   
   For actual functionality I tested all the basic druid functionality 
(ingesting data from azure, creating and loading segments, managing task logs, 
killing segments), but I didn't test any of the retry logic that the new 
clients promise.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to