[GLUTEN-2638][VL]fix s3 endpoint configuration (apache#2643)

set s3 endpoint only when not use instance credentials.

(Fixes: issue-2638)

cpp/velox/compute/VeloxInitializer.cc

@@ -171,9 +171,13 @@ void VeloxInitializer::init(const std::unordered_map<std::string, std::string>&
         {"hive.s3.aws-secret-key", awsSecretKey},
     });
   }
-
+  // Only need to set s3 endpoint when not use instance credentials.
+  if (useInstanceCredentials != "true") {
+    s3Config.insert({
+        {"hive.s3.endpoint", awsEndpoint},
+    });
+  }
   s3Config.insert({
-      {"hive.s3.endpoint", awsEndpoint},
       {"hive.s3.ssl.enabled", sslEnabled},
       {"hive.s3.path-style-access", pathStyleAccess},
   });

docs/get-started/VeloxS3.md

@@ -31,6 +31,7 @@ S3 also provides other methods for accessing, you can also use instance credenti
 ```
 spark.hadoop.fs.s3a.use.instance.credentials true
 ```
+Note that in this case, "spark.hadoop.fs.s3a.endpoint" won't take affect as Gluten will use the endpoint set during instance creation.
 
 ## Configuring S3 IAM roles
 You can also use iam role credentials by setting the following configurations. Instance credentials have higher priority than iam credentials.
@@ -60,4 +61,4 @@ spark.gluten.sql.columnar.backend.velox.ssdCacheIOThreads // the IO threads for
 spark.gluten.sql.columnar.backend.velox.ssdODirect        // enbale or disable O_DIRECT on cache write, default false.
 ```
 
-It's recommended to mount SSDs to the cache path to get the best performance of local caching. On the start up of Spark context, the cache files will be allocated under "spark.gluten.sql.columnar.backend.velox.cachePath", with UUID based suffix, e.g. "/tmp/cache.13e8ab65-3af4-46ac-8d28-ff99b2a9ec9b0". Gluten is not able to reuse older caches for now, and the old cache files are left there after Spark context shutdown.
+It's recommended to mount SSDs to the cache path to get the best performance of local caching. On the start up of Spark context, the cache files will be allocated under "spark.gluten.sql.columnar.backend.velox.cachePath", with UUID based suffix, e.g. "/tmp/cache.13e8ab65-3af4-46ac-8d28-ff99b2a9ec9b0". Gluten is not able to reuse older caches for now, and the old cache files are left there after Spark context shutdown.