|
| 1 | += Using the Azure Data Lake Storage (ADLS) |
| 2 | +:description: Use the Azure Data Lake Storage (ADLS) instead of HDFS for HBase. |
| 3 | +:abfs-authentication: https://hadoop.apache.org/docs/stable/hadoop-azure/abfs.html#Authentication |
| 4 | +:hadoop-18516: https://issues.apache.org/jira/browse/HADOOP-18516 |
| 5 | + |
| 6 | +Instead of HDFS, the HBase data can be stored in the Azure Data Lake Storage (ADLS). |
| 7 | +HDFS is then not required anymore to run HBase. |
| 8 | + |
| 9 | +A custom ConfigMap for ADLS containing the core-site.xml and an empty hdfs-site.xml must be provided: |
| 10 | + |
| 11 | +[source,yaml] |
| 12 | +---- |
| 13 | +--- |
| 14 | +apiVersion: v1 |
| 15 | +kind: ConfigMap |
| 16 | +metadata: |
| 17 | + name: adls |
| 18 | +data: |
| 19 | + core-site.xml: |- |
| 20 | + <?xml version="1.0"?> |
| 21 | + <configuration> |
| 22 | + <property> |
| 23 | + <name>fs.defaultFS</name> <!--1--> |
| 24 | + <value>abfs://<container-name>@<storage-account>.dfs.core.windows.net/</value> |
| 25 | + </property> |
| 26 | + <property> |
| 27 | + <name>fs.azure.account.auth.type.<storage-account>.dfs.core.windows.net</name> <!--2--> |
| 28 | + <value>SAS</value> |
| 29 | + </property> |
| 30 | + <property> |
| 31 | + <name>fs.azure.sas.fixed.token.<storage-account>.dfs.core.windows.net</name> <!--3--> |
| 32 | + <value>${env.SAS_TOKEN}</value> |
| 33 | + </property> |
| 34 | + <!-- Add further properties, e.g. for Kerberos. --> |
| 35 | + </configuration> |
| 36 | + hdfs-site.xml: |- |
| 37 | + <?xml version="1.0"?> |
| 38 | + <configuration> |
| 39 | + </configuration> |
| 40 | +---- |
| 41 | +<1> The name of the default file system. |
| 42 | +The `hadoop-azure` module which provides support for ADLS through the "abfs" connector, is already contained in the Stackable HBase image and accessible on the classpath. |
| 43 | +Make sure that the hierarchical namespace is enabled in the storage account. |
| 44 | +Replace `<container-name>` and `<storage-account>` accordingly. |
| 45 | +<2> The authentication mechanism. |
| 46 | +Possible values are `SharedKey`, `OAuth`, `Custom` and `SAS`, see the {abfs-authentication}[documentation] of the `hadoop-azure` module for further information. |
| 47 | +<3> If the authentication type is set to `SAS` (shared access signature) and no SAS token provider is given in `fs.azure.sas.token.provider.type`, then a fixed SAS token can be used instead. |
| 48 | +The `FixedSASTokenProvider` (see {hadoop-18516}[HADOOP-18516]) is back-ported to all supported HBase versions, so that the SAS token can be configured with this property. |
| 49 | + |
| 50 | +The ConfigMap must be referenced in `hdfsConfigMapName`. |
| 51 | +In the example above, the SAS token is read from an environment variable which is taken from a Secret: |
| 52 | + |
| 53 | +[source,yaml] |
| 54 | +---- |
| 55 | +--- |
| 56 | +apiVersion: hbase.stackable.tech/v1alpha1 |
| 57 | +kind: HbaseCluster |
| 58 | +spec: |
| 59 | + clusterConfig: |
| 60 | + hdfsConfigMapName: adls |
| 61 | + masters: |
| 62 | + podOverrides: |
| 63 | + spec: |
| 64 | + containers: |
| 65 | + - name: hbase |
| 66 | + env: |
| 67 | + - name: SAS_TOKEN |
| 68 | + valueFrom: |
| 69 | + secretKeyRef: |
| 70 | + name: adls-credentials |
| 71 | + key: sas-token |
| 72 | + regionServers: |
| 73 | + podOverrides: |
| 74 | + spec: |
| 75 | + containers: |
| 76 | + - name: hbase |
| 77 | + env: |
| 78 | + - name: SAS_TOKEN |
| 79 | + valueFrom: |
| 80 | + secretKeyRef: |
| 81 | + name: adls-credentials |
| 82 | + key: sas-token |
| 83 | +--- |
| 84 | +apiVersion: v1 |
| 85 | +kind: Secret |
| 86 | +metadata: |
| 87 | + name: adls-credentials |
| 88 | +type: Opaque |
| 89 | +data: |
| 90 | + # decoded sas-token: sp=racwdlmeop&st=2024-01-01T00:00:00Z&se=2025-01-01T00:00:00Z&spr=https&sv=2022-11-02&sr=c&sig=xxx |
| 91 | + sas-token: c3A9cmFjd2RsbWVvcCZzdD0yMDI0LTAxLTAxVDAwOjAwOjAwWiZzZT0yMDI1LTAxLTAxVDAwOjAwOjAwWiZzcHI9aHR0cHMmc3Y9MjAyMi0xMS0wMiZzcj1jJnNpZz14eHgK |
| 92 | +---- |
0 commit comments