Azure storage auth issue via oauth in tensorflow: Problem with the SSL CA cert (path? access rights?)

Question

I am trying to access container via oauth token in azure blob storage with tensorflow

The following python code works

from azure.storage.blob import BlobClient, ContainerClient

url = "https://{}.blob.core.windows.net".format(
    "my_storage_name"
)


from azure.identity import ClientSecretCredential
token_credential = ClientSecretCredential(
    "tenant_id,
    "ad_application_id",
    "ad_application_secret" 
)

blob_client = ContainerClient(account_url=oauth_url, container_name="container_name", credential=token_credential)

bloblist = blob_client.list_blobs()

for x in bloblist:
    print(x)

However when I tried to hack this in tensorflow io

std::string tenantId = "";
std::string activeDirectoryApplicationId = "";
std::string activeDirectoryApplicationSecret = "";

auto clientCertificateCredential
    = std::make_shared<Azure::Identity::ClientSecretCredential>(tenantId, activeDirectoryApplicationId, activeDirectoryApplicationSecret);

std::string storageContainerUrl = "https://mystorage.blob.core.windows.net/mycontainer";

auto blobclient = std::make_shared<Azure::Storage::Blobs::BlobContainerClient>(storageContainerUrl, clientCertificateCredential);

for (auto blobPage = (*blobclient).ListBlobs();
  blobPage.HasPage();
  blobPage.MoveToNextPage())
{
for (auto& blob : blobPage.Blobs)
{
 // Below is what you want to do with each blob
    TF_VLog(3, "blob: %s\n", blob.Name);
}}

I run into

RuntimeError: Fail to get a new connection for: httpslogin.microsoftonline.com. Problem with the SSL CA cert (path? access rights?)

even after changing to http in url

os.environ['TF_AZURE_STORAGE_USE_HTTP'] = '1'

I tried locally with the c++ code as well without tensorflow io, and it works, is there special setting within tensorflow I need to change?

Answer 1

• if you have set the TF_AZURE_STORAGE_USE_HTTP to '1', then you are using local emulator ie azurite

• To use the azure blob storage you will have to set it to the key which is available in portal.

• you can also use sas tokens to access blob storage. The sas tokens can be generated from the portal

• The process of accessing the blob using key and sas is similar.

Refer this tensorflow docs regarding blob storage.