Endpoint URLs for Blob and Dfs

[mkomadel]

Problem description
Hi, in our usecase we need to send data from our on premise java services to azure storage account. We do this using DataLakeServiceClient from your java SDK. Because of security reasons our outbound traffic has to go over proxy URL. We provide this URL simply using
DataLakeServiceClient serviceClient = new DataLakeServiceClientBuilder()
.endpoint('https://proxy.url')
.credential(....)
.buildClient();

The problem is, your SDK uses in fact 2 URLs. One for dfs, one for blob and unfortunately it simply replaces '.dfs.' occurence with '.blob.'

https://github.com/Azure/azure-sdk-for-java/blob/main/sdk/storage/azure-storage-file-datalake/src/main/java/com/azure/storage/file/datalake/DataLakeServiceClientBuilder.java#L162-L163

Our proxy domains are not very configurable and thus do not contain '.dfs.' subdomain. Additionally routing based on HTTP URL context is also not possible.

Proposed solution
We would like to have the possibility to set the URLs for dfs and blob endpoints separately

[github-actions]

Thank you for your feedback. Tagging and routing to the team member(s) best able to assist.

//cc: @browndav-msft @gunjansingh-msft @ibrandes @kyleknap @seanmcc-msft

[github-actions]

🎯 Agentic Issue Triage

Summary: The author needs a way to configure separate endpoint URLs for the DFS and Blob endpoints in DataLakeServiceClient, because the current SDK derives the blob URL by replacing .dfs. with .blob. in the provided endpoint — which doesn't work when routing through a proxy that doesn't have .dfs. in its domain.

📋 Issue Details

Package: com.azure:azure-storage-file-datalake
Affected API: DataLakeServiceClientBuilder.endpoint() / internal URL construction (line 162-163 of DataLakeServiceClientBuilder.java)
Scenarios:

• On-premise Java services sending data to Azure Storage via a proxy URL
• Proxy URL does not contain .dfs. subdomain, so the SDK's automatic blob URL derivation fails

Root ask: Provide an API to set the DFS and Blob endpoint URLs independently in DataLakeServiceClientBuilder, rather than deriving the blob URL by string replacement on the DFS URL.

🔎 Debugging / Reproduction Notes

The SDK constructs the blob service URL from the DFS endpoint by replacing .dfs. with .blob. (see DataLakeServiceClientBuilder.java lines 162-163). When using a proxy endpoint that doesn't include .dfs. in its hostname, this substitution either produces an incorrect URL or fails entirely.

Suggested investigation steps:

1. Confirm current behavior in DataLakeServiceClientBuilder — how blob URL is derived from the DFS endpoint
2. Evaluate whether a new builder method (e.g., blobServiceEndpoint(String url)) could allow explicit override of the blob endpoint
3. Review impact on other DataLake*ClientBuilder classes that may share the same pattern
4. Consider whether an HttpPipelinePolicy or ProxyOptions workaround exists today that could serve as interim guidance for the user

🏷️ Label Confidence

• Category: Client — The issue is about SDK behavior (URL construction logic in the builder), not the Azure service REST API itself
• Service: Storage — The affected package is azure-storage-file-datalake under /sdk/storage/
• Confidence: High — Package, service directory, and affected class are explicitly named in the issue body

👥 Owner Routing

• Matched CODEOWNERS entry: # ServiceLabel: %Storage (line 482) — issue's "Storage" label matches
• AzureSdkOwners: @browndav-msft @gunjansingh-msft @ibrandes @kyleknap @seanmcc-msft
• ServiceOwners: @xgithubtriage
• Routing action: Assigned @browndav-msft (randomly selected from AzureSdkOwners); added needs-team-attention
• Scan notes: No more specific multi-label entry for Storage+Client exists in CODEOWNERS; %Storage catch-all matched first during bottom-to-top scan

Generated by Agentic Triage for #49781 · 28.9 AIC · ⌖ 7.13 AIC · ⊞ 11K · ◷

[ibrandes]

Hi @mkomadel,

Does your proxy support HTTP CONNECT tunneling for HTTPS traffic? If yes, you could use the ProxyOptions class to configure your endpoint:

ProxyOptions proxyOptions = new ProxyOptions(
    ProxyOptions.Type.HTTP,
    new InetSocketAddress("proxy.company.com", 8080)
);

DataLakeServiceClient serviceClient = new DataLakeServiceClientBuilder()
    .endpoint("https://myaccount.dfs.core.windows.net") // real Azure URL
    .credential(credential)
    .clientOptions(new HttpClientOptions().setProxyOptions(proxyOptions))
    .buildClient();

Additionally, there's ProxyOptions.fromConfiguration(configuration) which automatically reads standard environment variables (HTTPS_PROXY, HTTP_PROXY) or Java system properties (https.proxyHost, https.proxyPort).