[
https://issues.apache.org/jira/browse/HADOOP-19379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17910005#comment-17910005
]
ASF GitHub Bot commented on HADOOP-19379:
-----------------------------------------
anujmodi2021 commented on PR #7252:
URL: https://github.com/apache/hadoop/pull/7252#issuecomment-2572328303
Thanks for the PR @horizonzy
We have a few things that we want to understand better. Can you please
update the description with proper explanation of requirement and how this code
change helping?
1. Why was this change made on wasb driver? [Wasb is on a path of
deprecation](https://issues.apache.org/jira/browse/HADOOP-19178) and we already
support
2. Why do we need any other URL when Azurite is supposed to run on localhost
always?
3. What does "another container" means here? Do you want to work with
multiple filesystems?
@steveloughran we already have support for azurite emulator on ABFS. It
might need better documentation though, will chcek and work on that. But for CI
and PR sign offs we would still prefer testing using proper production accounts
due to following reasons.
1. Azure storage services are rapidly changing including/updating a lot of
features. We are not sure how often azurite is updated to reflect those changes.
2. Any release and update of azurite will force us to update the CI setup,
that will be an additional manual maintainence.
There is a value in terms of saving time during local dev and there we can
consider using this more often.
> [ABFS] Support Azurite storage emulator
> ---------------------------------------
>
> Key: HADOOP-19379
> URL: https://issues.apache.org/jira/browse/HADOOP-19379
> Project: Hadoop Common
> Issue Type: Improvement
> Components: fs/azure, test
> Affects Versions: 3.4.1
> Reporter: Yan Zhao
> Priority: Minor
> Labels: pull-request-available
>
> In the integration test case, we will start the Azurite using the test
> container, and we need to access Azurite at another container.
> Now, the Azurite emulator uri values is fixed at 127.0.0.1, so at another
> container, it can't access the Azurite service.
> So I want to introduce a new config `fs.azure.storage.emulator.proxy.url`, so
> that I can config `fs.azure.storage.emulator.proxy.url=http://\{AzuriteIp}`,
> then it can access the Azurite service.
>
>
> {code:java}
> private void connectUsingCredentials(String accountName,
> StorageCredentials credentials, String containerName)
> throws URISyntaxException, StorageException, AzureException {
> URI blobEndPoint;
> if (isStorageEmulatorAccount(accountName)) {
> isStorageEmulator = true;
> CloudStorageAccount account =
> CloudStorageAccount.getDevelopmentStorageAccount();
> storageInteractionLayer.createBlobClient(account);
> } else {
> blobEndPoint = new URI(getHTTPScheme() + "://" + accountName);
> storageInteractionLayer.createBlobClient(blobEndPoint, credentials);
> }
> suppressRetryPolicyInClientIfNeeded();
> // Capture the container reference for debugging purposes.
> container = storageInteractionLayer.getContainerReference(containerName);
> rootDirectory = container.getDirectoryReference("");
> // Can only create container if using account key credentials
> canCreateOrModifyContainer = credentials instanceof
> StorageCredentialsAccountAndKey;
> } {code}
>
> {code:java}
> public static CloudStorageAccount getDevelopmentStorageAccount() {
> try {
> return getDevelopmentStorageAccount(null);
> }
> catch (final URISyntaxException e) {
> // this won't happen since we know the standard development stororage
> uri is valid.
> return null;
> }
> }
> public static CloudStorageAccount getDevelopmentStorageAccount(final URI
> proxyUri) throws URISyntaxException {
> String scheme;
> String host;
> if (proxyUri == null) {
> scheme = "http";
> host = "127.0.0.1";
> }
> else {
> scheme = proxyUri.getScheme();
> host = proxyUri.getHost();
> }
> StorageCredentials credentials = new
> StorageCredentialsAccountAndKey(DEVSTORE_ACCOUNT_NAME,
> DEVSTORE_ACCOUNT_KEY);
> URI blobPrimaryEndpoint = new
> URI(String.format(DEVELOPMENT_STORAGE_PRIMARY_ENDPOINT_FORMAT, scheme, host,
> "10000", DEVSTORE_ACCOUNT_NAME));
> URI queuePrimaryEndpoint = new
> URI(String.format(DEVELOPMENT_STORAGE_PRIMARY_ENDPOINT_FORMAT, scheme, host,
> "10001", DEVSTORE_ACCOUNT_NAME));
> URI tablePrimaryEndpoint = new
> URI(String.format(DEVELOPMENT_STORAGE_PRIMARY_ENDPOINT_FORMAT, scheme, host,
> "10002", DEVSTORE_ACCOUNT_NAME));
> URI blobSecondaryEndpoint = new
> URI(String.format(DEVELOPMENT_STORAGE_SECONDARY_ENDPOINT_FORMAT, scheme, host,
> "10000", DEVSTORE_ACCOUNT_NAME));
> URI queueSecondaryEndpoint = new
> URI(String.format(DEVELOPMENT_STORAGE_SECONDARY_ENDPOINT_FORMAT, scheme, host,
> "10001", DEVSTORE_ACCOUNT_NAME));
> URI tableSecondaryEndpoint = new
> URI(String.format(DEVELOPMENT_STORAGE_SECONDARY_ENDPOINT_FORMAT, scheme, host,
> "10002", DEVSTORE_ACCOUNT_NAME));
> CloudStorageAccount account = new CloudStorageAccount(credentials, new
> StorageUri(blobPrimaryEndpoint,
> blobSecondaryEndpoint), new StorageUri(queuePrimaryEndpoint,
> queueSecondaryEndpoint), new StorageUri(
> tablePrimaryEndpoint, tableSecondaryEndpoint), null /*
> fileStorageUri */);
> account.isDevStoreAccount = true;
> return account;
> } {code}
> CloudStorageAccount account =
> CloudStorageAccount.getDevelopmentStorageAccount(); will using 127.0.0.1 as
> the `azurite` host.
>
> In fact, here we can pass into a proxy uri by invoke
> getDevelopmentStorageAccount(final URI proxyUri)
>
>
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
