|
| 1 | +<!-- |
| 2 | + - Licensed to the Apache Software Foundation (ASF) under one or more |
| 3 | + - contributor license agreements. See the NOTICE file distributed with |
| 4 | + - this work for additional information regarding copyright ownership. |
| 5 | + - The ASF licenses this file to You under the Apache License, Version 2.0 |
| 6 | + - (the "License"); you may not use this file except in compliance with |
| 7 | + - the License. You may obtain a copy of the License at |
| 8 | + - |
| 9 | + - http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | + - |
| 11 | + - Unless required by applicable law or agreed to in writing, software |
| 12 | + - distributed under the License is distributed on an "AS IS" BASIS, |
| 13 | + - WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | + - See the License for the specific language governing permissions and |
| 15 | + - limitations under the License. |
| 16 | + --> |
| 17 | + |
| 18 | +<div align=center> |
| 19 | + |
| 20 | + |
| 21 | + |
| 22 | +</div> |
| 23 | + |
| 24 | +# Access Kerberized Kyuubi with Beeline & BI Tools |
| 25 | + |
| 26 | +## Instructions |
| 27 | +When Kyuubi is secured by Kerberos, the authentication procedure becomes a little complicated. |
| 28 | + |
| 29 | + |
| 30 | + |
| 31 | +The graph above shows a simplified kerberos authentication procedure: |
| 32 | +1. Kerberos client sends user principal and secret key to KDC. Secret key can be a password or a keytab file. |
| 33 | +2. KDC returns a `ticket-granting ticket`(TGT). |
| 34 | +3. Kerberos client stores TGT into a ticket cache. |
| 35 | +4. JDBC client, such as beeline and BI tools, reads TGT from the ticket cache. |
| 36 | +5. JDBC client sends TGT and server principal to KDC. |
| 37 | +6. KDC returns a `client-to-server ticket`. |
| 38 | +7. JDBC client sends `client-to-server ticket` to Kyuubi server to prove its identity. |
| 39 | + |
| 40 | +In the rest part of this page, we will describe steps needed to pass through this authentication. |
| 41 | + |
| 42 | +## Install Kerberos Client |
| 43 | +Usually, Kerberos client is installed as default. You can validate it using klist tool. |
| 44 | + |
| 45 | +Linux command and output: |
| 46 | +```bash |
| 47 | +$ klist -V |
| 48 | +Kerberos 5 version 1.15.1 |
| 49 | +``` |
| 50 | + |
| 51 | +MacOS command and output: |
| 52 | +```bash |
| 53 | +$ klist --version |
| 54 | +klist (Heimdal 1.5.1apple1) |
| 55 | +Copyright 1995-2011 Kungliga Tekniska Högskolan |
| 56 | +Send bug-reports to heimdal-bugs@h5l.org |
| 57 | +``` |
| 58 | + |
| 59 | +Windows command and output: |
| 60 | +```cmd |
| 61 | +> klist -V |
| 62 | +Kerberos for Windows |
| 63 | +``` |
| 64 | + |
| 65 | +If the client is not installed, you should install it ahead based on the OS platform. |
| 66 | +We recommend you to install the MIT Kerberos Distribution as all commands in this guide is based on it. |
| 67 | + |
| 68 | +## Configure Kerberos Client |
| 69 | +Kerberos client needs a configuration file for tuning up the creation of Kerberos ticket cache. |
| 70 | +Following is the configuration file's default location on different OS: |
| 71 | + |
| 72 | +OS | Path |
| 73 | +---| --- |
| 74 | +Linux | /etc/krb5.conf |
| 75 | +MacOS | /etc/krb5.conf |
| 76 | +Windows | %ProgramData%\MIT\Kerberos5\krb5.ini |
| 77 | + |
| 78 | +You can use `KRB5_CONFIG` environment variable to overwrite the default location. |
| 79 | + |
| 80 | +The configuration file should be configured to point to the same KDC as Kyuubi points to. |
| 81 | + |
| 82 | +## Get Kerberos TGT |
| 83 | +Execute `kinit` command to get TGT from KDC. |
| 84 | + |
| 85 | +Suppose user principal is `kyuubi_user@KYUUBI.APACHE.ORG` and user keytab file name is `kyuubi_user.keytab`, |
| 86 | +the command should be: |
| 87 | + |
| 88 | +``` |
| 89 | +$ kinit -kt kyuubi_user.keytab kyuubi_user@KYUUBI.APACHE.ORG |
| 90 | +
|
| 91 | +(Command is identical on different OS platform) |
| 92 | +``` |
| 93 | + |
| 94 | +You may also execute `kinit` command with principal and password to get TGT: |
| 95 | + |
| 96 | +``` |
| 97 | +$ kinit kyuubi_user@KYUUBI.APACHE.ORG |
| 98 | +Password for kyuubi_user@KYUUBI.APACHE.ORG: password |
| 99 | +
|
| 100 | +(Command is identical on different OS platform) |
| 101 | +``` |
| 102 | + |
| 103 | +If the command executes successfully, TGT will be store in ticket cache. |
| 104 | +Use `klist` command to print TGT info in ticket cache: |
| 105 | + |
| 106 | +``` |
| 107 | +$ klist |
| 108 | +
|
| 109 | +Ticket cache: FILE:/tmp/krb5cc_1000 |
| 110 | +Default principal: kyuubi_user@KYUUBI.APACHE.ORG |
| 111 | +
|
| 112 | +Valid starting Expires Service principal |
| 113 | +2021-12-13T18:44:58 2021-12-14T04:44:58 krbtgt/KYUUBI.APACHE.ORG@KYUUBI.APACHE.ORG |
| 114 | + renew until 2021-12-14T18:44:57 |
| 115 | + |
| 116 | +(Command is identical on different OS platform. Ticket cache location may be different.) |
| 117 | +``` |
| 118 | + |
| 119 | +Ticket cache may have different storage type on different OS platform. |
| 120 | + |
| 121 | +For example, |
| 122 | + |
| 123 | +OS | Default Ticket Cache Type and Location |
| 124 | +---| --- |
| 125 | +Linux | FILE:/tmp/krb5cc_%{uid} |
| 126 | +MacOS | KCM:%{uid}:%{gid} |
| 127 | +Windows | API:krb5cc |
| 128 | + |
| 129 | +You can find your ticket cache type and location in the `Ticket cache` part of `klist` output. |
| 130 | + |
| 131 | +**Note**: |
| 132 | +- Ensure your ticket cache type is `FILE` as JVM can only read ticket cache stored as file. |
| 133 | +- Do not store TGT into default ticket cache if you are running Kyuubi and execute `kinit` on the same |
| 134 | +host with the same OS user. The default ticket cache is already used by Kyuubi server. |
| 135 | + |
| 136 | +Either because the default ticket cache is not a file, or because it is used by Kyuubi server, you |
| 137 | +should store ticket cache in another file location. |
| 138 | +This can be achieved by specifying a file location with `-c` argument in `kinit` command. |
| 139 | + |
| 140 | +For example, |
| 141 | +``` |
| 142 | +$ kinit -c /tmp/krb5cc_beeline -kt kyuubi_user.keytab kyuubi_user@KYUUBI.APACHE.ORG |
| 143 | +
|
| 144 | +(Command is identical on different OS platform) |
| 145 | +``` |
| 146 | + |
| 147 | +To check the ticket cache, specify the file location with `-c` argument in `klist` command. |
| 148 | + |
| 149 | +For example, |
| 150 | +``` |
| 151 | +$ klist -c /tmp/krb5cc_beeline |
| 152 | +
|
| 153 | +(Command is identical on different OS platform) |
| 154 | +``` |
| 155 | + |
| 156 | +## Add Kerberos Client Configuration File to JVM Search Path |
| 157 | +The JVM, which JDBC client is running on, also needs to read the Kerberos client configuration file. |
| 158 | +However, JVM uses different default locations from Kerberos client, and does not honour `KRB5_CONFIG` |
| 159 | +environment variable. |
| 160 | + |
| 161 | +OS | JVM Search Paths |
| 162 | +---| --- |
| 163 | +Linux | System scope: `/etc/krb5.conf` |
| 164 | +MacOS | User scope: `$HOME/Library/Preferences/edu.mit.Kerberos`<br/>System scope: `/etc/krb5.conf` |
| 165 | +Windows | User scoep: `%USERPROFILE%\krb5.ini`<br/>System scope: `%windir%\krb5.ini` |
| 166 | + |
| 167 | +You can use JVM system property, `java.security.krb5.conf`, to overwrite the default location. |
| 168 | + |
| 169 | +## Add Kerberos Ticket Cache to JVM Search Path |
| 170 | +JVM determines the ticket cache location in the following order: |
| 171 | +1. Path specified by `KRB5CCNAME` environment variable. Path must start with `FILE:`. |
| 172 | +2. `/tmp/krb5cc_%{uid}` on Unix-like OS, e.g. Linux, MacOS |
| 173 | +3. `${user.home}/krb5cc_${user.name}` if `${user.name}` is not null |
| 174 | +4. `${user.home}/krb5cc` if `${user.name}` is null |
| 175 | + |
| 176 | +**Note**: |
| 177 | +- `${user.home}` and `${user.name}` are JVM system properties. |
| 178 | +- `${user.home}` should be replaced with `${user.dir}` if `${user.home}` is null. |
| 179 | + |
| 180 | +Ensure your ticket cache is stored as a file and put it in one of the above locations. |
| 181 | + |
| 182 | +## Ensure core-site.xml Exists in Classpath |
| 183 | +Like hadoop clients, `hadoop.security.authentication` should be set to `KERBEROS` in `core-site.xml` |
| 184 | +to let Hive JDBC driver use Kerberos authentication. `core-site.xml` should be placed under beeline's |
| 185 | +classpath or BI tools' classpath. |
| 186 | + |
| 187 | +### Beeline |
| 188 | +Here are the usual locations where `core-site.xml` should exist for different beeline distributions: |
| 189 | + |
| 190 | +Client | Location | Note |
| 191 | +--- | --- | --- |
| 192 | +Hive beeline | `$HADOOP_HOME/etc/hadoop` | Hive resolves `$HADOOP_HOME` and use `$HADOOP_HOME/bin/hadoop` command to launch beeline. `$HADOOP_HOME/etc/hadoop` is in `hadoop` command's classpath. |
| 193 | +Spark beeline | `$HADOOP_CONF_DIR` | In `$SPARK_HOME/conf/spark-env.sh`, `$HADOOP_CONF_DIR` often be set to the directory containing hadoop client configuration files. |
| 194 | +Kyuubi beeline | `$HADOOP_CONF_DIR` | In `$KYUUBI_HOME/conf/kyuubi-env.sh`, `$HADOOP_CONF_DIR` often be set to the directory containing hadoop client configuration files. |
| 195 | + |
| 196 | +If `core-site.xml` is not found in above locations, create one with the following content: |
| 197 | +```xml |
| 198 | +<configuration> |
| 199 | + <property> |
| 200 | + <name>hadoop.security.authentication</name> |
| 201 | + <value>kerberos</value> |
| 202 | + </property> |
| 203 | +</configuration> |
| 204 | +``` |
| 205 | + |
| 206 | +### BI Tools |
| 207 | +As to BI tools, ways to add `core-site.xml` varies. |
| 208 | +Take DBeaver as an example. We can add files to DBeaver's classpath through its `Global libraries` preference. |
| 209 | +As `Global libraries` only accepts jar files, you should package `core-site.xml` into a jar file. |
| 210 | + |
| 211 | +```bash |
| 212 | +$ jar -c -f core-site.jar core-site.xml |
| 213 | + |
| 214 | +(Command is identical on different OS platform) |
| 215 | +``` |
| 216 | + |
| 217 | +## Connect with JDBC URL |
| 218 | +The last step is to connect to Kyuubi with the right JDBC URL. |
| 219 | +The JDBC URL should be in format: |
| 220 | + |
| 221 | +``` |
| 222 | +jdbc:hive2://<kyuubi_server_a ddress>:<kyuubi_server_port>/<db>;principal=<kyuubi_server_principal> |
| 223 | +``` |
| 224 | + |
| 225 | +**Note**: |
| 226 | +- `kyuubi_server_principal` is the value of `kyuubi.kinit.principal` set in `kyuubi-defaults.conf`. |
| 227 | +- As a command line argument, JDBC URL should be quoted to avoid being split into 2 commands by ";". |
| 228 | +- As to DBeaver, `<db>;principal=<kyuubi_server_principal>` should be set as the `Database/Schema` argument. |
| 229 | + |
0 commit comments