瀏覽代碼

HADOOP-19404: [ABFS][FNSOverBlob] Update documentation for FNS Blob Onboard (#7341) (#7398)

Contributed by Anmol Asrani
Reviewed and Signed off by Anuj Modi
Anmol Asrani 3 月之前
父節點
當前提交
fc388f782c
共有 1 個文件被更改,包括 77 次插入3 次删除
  1. 77 3
      hadoop-tools/hadoop-azure/src/site/markdown/fns_blob.md

+ 77 - 3
hadoop-tools/hadoop-azure/src/site/markdown/fns_blob.md

@@ -14,8 +14,6 @@
 
 # ABFS Driver for Namespace Disabled Accounts (FNS: Flat Namespace)
 
-### Note: FNS-BLOB Support is being built and not yet ready for usage.
-
 ## Background
 The ABFS driver is recommended to be used only with HNS Enabled ADLS Gen-2 accounts
 for big data analytics because of being more performant and scalable.
@@ -79,4 +77,80 @@ configured service type.
         <name>fs.azure.fns.account.service.type</name>
         <value>BLOB</value>
     </property>
-    ```
+    ```
+
+5. Auth type supported over FNS Blob: SharedKey, OAuth and fixed SAS.
+    ```xml
+   <property>
+        <name>fs.azure.account.auth.type</name>
+        <value>SharedKey</value>
+    </property>
+    ```
+    - How to configure Shared Key
+      auth: [Shared Key](./abfs.md#a-nameshared-key-autha-default-shared-key)
+    - How to configure
+      OAuth: [OAuth](./abfs.md#a-nameoauth-client-credentialsa-oauth-20-client-credentials)
+    - How to configure fixed
+      SAS: [Fixed SAS](./abfs.md#using-accountservice-sas-with-abfs)
+
+   OAuth is recommended auth type as it is more secure and flexible.
+
+### <a name="renamedeleteoptions"></a> Rename delete configs
+
+The following configs are related to rename and delete operations.
+
+- `fs.azure.blob.copy.progress.wait.millis`: Blob copy API is an async API, this
+  configuration defines polling duration for checking copy status. The default
+  value is 1 sec i.e. 1000 ms.
+
+- `fs.azure.blob.copy.max.wait.millis`: Maximum time to wait for a blob copy
+  operation to complete. The default value is 5 minutes.
+
+- `fs.azure.blob.atomic.rename.lease.refresh.duration`: Blob rename lease
+  refresh
+  duration in milliseconds. This setting ensures that the lease on the blob is
+  periodically refreshed during a rename operation to prevent other operations
+  from interfering.
+  The default value is 60 seconds.
+
+- `fs.azure.blob.dir.list.producer.queue.max.size`: Maximum number of blob
+  entries
+  enqueued in memory for rename or delete orchestration. The default value is 2
+  times the default value of list max results, which is 5000, making the current
+  value 10000.
+
+- `fs.azure.blob.dir.list.consumer.max.lag`: It sets a limit on how much blob
+  information can be waiting to be processed (consumer lag) during a blob
+  listing
+  operation. If the amount of unprocessed blob information exceeds this limit,
+  the
+  producer will pause until the consumer catches up and the lag becomes
+  manageable. The default value is equal to the value of default value of list
+  max
+  results which is 5000 currently.
+
+- `fs.azure.blob.dir.rename.max.thread`: Maximum number of threads per blob
+  rename
+  orchestration. The default value is 5.
+
+- `fs.azure.blob.dir.delete.max.thread`: Maximum number of thread per
+  blob-delete
+  orchestration. The default value currently is 5.
+
+## Features currently not supported
+
+1. **User Delegation SAS** feature is currently not supported but we
+   plan to bring support for it in the future.
+   Jira to track this
+   workitem : https://issues.apache.org/jira/browse/HADOOP-19406.
+
+
+2. **Context Provider Key (CPK)** support is currently not available. It refers to the ability to use a
+customer-provided encryption key to encrypt and decrypt data in Azure Blob
+Storage. This feature allows users to manage their own encryption keys,
+providing an additional layer of security and control over their data.
+
+## Ask all about ABFS Driver
+
+For any queries related to onboard to FNS Blob or anything related to ABFS
+Driver in general, kindly reach out to us at **askabfs@microsoft.com**.