Introduction to the Kyuubi Configurations System
Kyuubi provides several ways to configure the system and corresponding engines.
Environments
You can configure the environment variables in $KYUUBI_HOME/conf/kyuubi-env.sh, e.g, JAVA_HOME, then this java runtime will be used both for Kyuubi server instance and the applications it launches. You can also change the variable in the subprocess's env configuration file, e.g.$SPARK_HOME/conf/spark-env.sh to use more specific ENV for SQL engine applications.
#!/usr/bin/env bash # # Licensed to the Apache Software Foundation (ASF) under one or more # contributor license agreements. See the NOTICE file distributed with # this work for additional information regarding copyright ownership. # The ASF licenses this file to You under the Apache License, Version 2.0 # (the "License"); you may not use this file except in compliance with # the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # # - JAVA_HOME Java runtime to use. By default use "java" from PATH. # # # - KYUUBI_CONF_DIR Directory containing the Kyuubi configurations to use. # (Default: $KYUUBI_HOME/conf) # - KYUUBI_LOG_DIR Directory for Kyuubi server-side logs. # (Default: $KYUUBI_HOME/logs) # - KYUUBI_PID_DIR Directory stores the Kyuubi instance pid file. # (Default: $KYUUBI_HOME/pid) # - KYUUBI_MAX_LOG_FILES Maximum number of Kyuubi server logs can rotate to. # (Default: 5) # - KYUUBI_JAVA_OPTS JVM options for the Kyuubi server itself in the form "-Dx=y". # (Default: none). # - KYUUBI_NICENESS The scheduling priority for Kyuubi server. # (Default: 0) # - KYUUBI_WORK_DIR_ROOT Root directory for launching sql engine applications. # (Default: $KYUUBI_HOME/work) # - HADOOP_CONF_DIR Directory containing the Hadoop / YARN configuration to use. # # - SPARK_HOME Spark distribution which you would like to use in Kyuubi. # - SPARK_CONF_DIR Optional directory where the Spark configuration lives. # (Default: $SPARK_HOME/conf) # ## Examples ## # export JAVA_HOME=/usr/jdk64/jdk1.8.0_152 # export HADOOP_CONF_DIR=/usr/ndp/current/mapreduce_client/conf # export KYUUBI_JAVA_OPTS="-Xmx10g -XX:+UnlockDiagnosticVMOptions -XX:ParGCCardsPerStrideChunk=4096 -XX:+UseParNewGC -XX:+UseConcMarkSweepGC -XX:+CMSConcurrentMTEnabled -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -XX:+CMSClassUnloadingEnabled -XX:+CMSParallelRemarkEnabled -XX:+UseCondCardMark -XX:MaxDirectMemorySize=1024m -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=./logs -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps -XX:+PrintTenuringDistribution -Xloggc:./logs/kyuubi-server-gc-%t.log -XX:+UseGCLogFileRotation -XX:NumberOfGCLogFiles=10 -XX:GCLogFileSize=5M -XX:NewRatio=3 -XX:MetaspaceSize=512m"
Kyuubi Configurations
You can configure the Kyuubi properties in $KYUUBI_HOME/conf/kyuubi-defaults.conf. For example:
# # Licensed to the Apache Software Foundation (ASF) under one or more # contributor license agreements. See the NOTICE file distributed with # this work for additional information regarding copyright ownership. # The ASF licenses this file to You under the Apache License, Version 2.0 # (the "License"); you may not use this file except in compliance with # the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # ## Kyuubi Configurations # # kyuubi.authentication NONE # kyuubi.frontend.bind.host localhost # kyuubi.frontend.bind.port 10009 # # Details in https://kyuubi.readthedocs.io/en/latest/deployment/settings.html
Authentication
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.authentication | NONE | Client authentication types. NOSASL: raw transport. NONE: no authentication check. KERBEROS: Kerberos/GSSAPI authentication. LDAP: Lightweight Directory Access Protocol authentication. | string | 1.0.0 |
| kyuubi.authentication .ldap.base.dn | <undefined> | LDAP base DN. | string | 1.0.0 |
| kyuubi.authentication .ldap.domain | <undefined> | LDAP domain. | string | 1.0.0 |
| kyuubi.authentication .ldap.url | <undefined> | SPACE character separated LDAP connection URL(s). | string | 1.0.0 |
| kyuubi.authentication .sasl.qop | auth | Sasl QOP enable higher levels of protection for Kyuubi communication with clients. auth - authentication only (default) auth-int - authentication plus integrity protection auth-conf - authentication plus integrity and confidentiality protection. This is applicable only if Kyuubi is configured to use Kerberos authentication. | string | 1.0.0 |
Backend
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.backend.engine .exec.pool.keepalive .time | PT1M | Time(ms) that an idle async thread of the operation execution thread pool will wait for a new task to arrive before terminating in SQL engine applications | duration | 1.0.0 |
| kyuubi.backend.engine .exec.pool.shutdown .timeout | PT10S | Timeout(ms) for the operation execution thread pool to terminate in SQL engine applications | duration | 1.0.0 |
| kyuubi.backend.engine .exec.pool.size | 100 | Number of threads in the operation execution thread pool of SQL engine applications | int | 1.0.0 |
| kyuubi.backend.engine .exec.pool.wait.queue .size | 100 | Size of the wait queue for the operation execution thread pool in SQL engine applications | int | 1.0.0 |
| kyuubi.backend.server .exec.pool.keepalive .time | PT1M | Time(ms) that an idle async thread of the operation execution thread pool will wait for a new task to arrive before terminating in Kyuubi server | duration | 1.0.0 |
| kyuubi.backend.server .exec.pool.shutdown .timeout | PT10S | Timeout(ms) for the operation execution thread pool to terminate in Kyuubi server | duration | 1.0.0 |
| kyuubi.backend.server .exec.pool.size | 100 | Number of threads in the operation execution thread pool of Kyuubi server | int | 1.0.0 |
| kyuubi.backend.server .exec.pool.wait.queue .size | 100 | Size of the wait queue for the operation execution thread pool of Kyuubi server | int | 1.0.0 |
Delegation
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.delegation.key .update.interval | PT24H | unused yet | duration | 1.0.0 |
| kyuubi.delegation .token.gc.interval | PT1H | unused yet | duration | 1.0.0 |
| kyuubi.delegation .token.max.lifetime | PT168H | unused yet | duration | 1.0.0 |
| kyuubi.delegation .token.renew.interval | PT168H | unused yet | duration | 1.0.0 |
Engine
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.engine .deregister.exception .classes | A comma separated list of exception classes. If there is any exception thrown, whose class matches the specified classes, the engine would deregister itself. | seq | 1.2.0 | |
| kyuubi.engine .deregister.exception .messages | A comma separated list of exception messages. If there is any exception thrown, whose message or stacktrace matches the specified message list, the engine would deregister itself. | seq | 1.2.0 | |
| kyuubi.engine .deregister.exception .ttl | PT30M | Time to live(TTL) for exceptions pattern specified in kyuubi.engine.deregister.exception.classes and kyuubi.engine.deregister.exception.messages to deregister engines. Once the total error count hits the kyuubi.engine.deregister.job.max.failures within the TTL, an engine will deregister itself and wait for self-terminated. Otherwise, we suppose that the engine has recovered from temporary failures. | duration | 1.2.0 |
| kyuubi.engine .deregister.job.max .failures | 4 | Number of failures of job before deregistering the engine. | int | 1.2.0 |
| kyuubi.engine .initialize.sql | SHOW DATABASES | SemiColon-separated list of SQL statements to be initialized in the newly created engine before queries. | string | 1.2.0 |
| kyuubi.engine.share .level | USER | Engines will be shared in different levels, available configs are: CONNECTION: engine will not be shared but only used by the current client connection USER: engine will be shared by all sessions created by a unique username, see also kyuubi.engine.share.level.sub.domain SERVER: the App will be shared by Kyuubi servers | string | 1.2.0 |
| kyuubi.engine.share .level.sub.domain | <undefined> | Allow end-users to create a sub-domain for the share level of an engine. A sub-domain is a case-insensitive string values in ^[a-zA-Z_]{1,10}$ form. For example, for USER share level, an end-user can share a certain engine within a sub-domain, not for all of its clients. End-users are free to create multiple engines in the USER share level | string | 1.2.0 |
Frontend
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.frontend .backoff.slot.length | PT0.1S | Time to back off during login to the frontend service. | duration | 1.0.0 |
| kyuubi.frontend.bind .host | <undefined> | Hostname or IP of the machine on which to run the frontend service. | string | 1.0.0 |
| kyuubi.frontend.bind .port | 10009 | Port of the machine on which to run the frontend service. | int | 1.0.0 |
| kyuubi.frontend.login .timeout | PT20S | Timeout for Thrift clients during login to the frontend service. | duration | 1.0.0 |
| kyuubi.frontend.max .message.size | 104857600 | Maximum message size in bytes a Kyuubi server will accept. | int | 1.0.0 |
| kyuubi.frontend.max .worker.threads | 999 | Maximum number of threads in the of frontend worker thread pool for the frontend service | int | 1.0.0 |
| kyuubi.frontend.min .worker.threads | 9 | Minimum number of threads in the of frontend worker thread pool for the frontend service | int | 1.0.0 |
| kyuubi.frontend .worker.keepalive.time | PT1M | Keep-alive time (in milliseconds) for an idle worker thread | duration | 1.0.0 |
Ha
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.ha.zookeeper .acl.enabled | false | Set to true if the zookeeper ensemble is kerberized | boolean | 1.0.0 |
| kyuubi.ha.zookeeper .connection.base.retry .wait | 1000 | Initial amount of time to wait between retries to the zookeeper ensemble | int | 1.0.0 |
| kyuubi.ha.zookeeper .connection.max .retries | 3 | Max retry times for connecting to the zookeeper ensemble | int | 1.0.0 |
| kyuubi.ha.zookeeper .connection.max.retry .wait | 30000 | Max amount of time to wait between retries for BOUNDED_EXPONENTIAL_BACKOFF policy can reach, or max time until elapsed for UNTIL_ELAPSED policy to connect the zookeeper ensemble | int | 1.0.0 |
| kyuubi.ha.zookeeper .connection.retry .policy | EXPONENTIAL_BACKOFF | The retry policy for connecting to the zookeeper ensemble, all candidates are: ONE_TIME N_TIME EXPONENTIAL_BACKOFF BOUNDED_EXPONENTIAL_BACKOFF UNTIL_ELAPSED | string | 1.0.0 |
| kyuubi.ha.zookeeper .connection.timeout | 15000 | The timeout(ms) of creating the connection to the zookeeper ensemble | int | 1.0.0 |
| kyuubi.ha.zookeeper .namespace | kyuubi | The root directory for the service to deploy its instance uri. Additionally, it will creates a -[username] suffixed root directory for each application | string | 1.0.0 |
| kyuubi.ha.zookeeper .node.creation.timeout | PT2M | Timeout for creating zookeeper node | duration | 1.2.0 |
| kyuubi.ha.zookeeper .quorum | The connection string for the zookeeper ensemble | string | 1.0.0 | |
| kyuubi.ha.zookeeper .session.timeout | 60000 | The timeout(ms) of a connected session to be idled | int | 1.0.0 |
Kinit
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.kinit.interval | PT1H | How often will Kyuubi server run kinit -kt [keytab] [principal] to renew the local Kerberos credentials cache | duration | 1.0.0 |
| kyuubi.kinit.keytab | <undefined> | Location of Kyuubi server's keytab. | string | 1.0.0 |
| kyuubi.kinit.max .attempts | 10 | How many times will kinit process retry | int | 1.0.0 |
| kyuubi.kinit .principal | <undefined> | Name of the Kerberos principal. | string | 1.0.0 |
Metrics
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.metrics .console.interval | PT5S | How often should report metrics to console | duration | 1.2.0 |
| kyuubi.metrics .enabled | true | Set to true to enable kyuubi metrics system | boolean | 1.2.0 |
| kyuubi.metrics.json .interval | PT5S | How often should report metrics to json file | duration | 1.2.0 |
| kyuubi.metrics.json .location | metrics | Where the json metrics file located | string | 1.2.0 |
| kyuubi.metrics .prometheus.path | /metrics | URI context path of prometheus metrics HTTP server | string | 1.2.0 |
| kyuubi.metrics .prometheus.port | 10019 | Prometheus metrics HTTP server port | int | 1.2.0 |
| kyuubi.metrics .reporters | JSON | A comma separated list for all metrics reporters CONSOLE - ConsoleReporter which outputs measurements to CONSOLE periodically. JMX - JmxReporter which listens for new metrics and exposes them as MBeans. JSON - JsonReporter which outputs measurements to json file periodically. PROMETHEUS - PrometheusReporter which exposes metrics in prometheus format. SLF4J - Slf4jReporter which outputs measurements to system log periodically. | seq | 1.2.0 |
| kyuubi.metrics.slf4j .interval | PT5S | How often should report metrics to SLF4J logger | duration | 1.2.0 |
Operation
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.operation.idle .timeout | PT3H | Operation will be closed when it's not accessed for this duration of time | duration | 1.0.0 |
| kyuubi.operation .interrupt.on.cancel | true | When true, all running tasks will be interrupted if one cancels a query. When false, all running tasks will remain until finished. | boolean | 1.2.0 |
| kyuubi.operation .query.timeout | <undefined> | Timeout for query executions at server-side, take affect with client-side timeout(java.sql.Statement.setQueryTimeout) together, a running query will be cancelled automatically if timeout. It's off by default, which means only client-side take fully control whether the query should timeout or not. If set, client-side timeout capped at this point. To cancel the queries right away without waiting task to finish, consider enabling kyuubi.operation.interrupt.on.cancel together. | duration | 1.2.0 |
| kyuubi.operation .scheduler.pool | <undefined> | The scheduler pool of job. Note that, this config should be used after change Spark config spark.scheduler.mode=FAIR. | string | 1.1.1 |
| kyuubi.operation .status.polling .timeout | PT5S | Timeout(ms) for long polling asynchronous running sql query's status | duration | 1.0.0 |
Session
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.session.check .interval | PT5M | The check interval for session timeout. | duration | 1.0.0 |
| kyuubi.session.conf .ignore.list | A comma separated list of ignored keys. If the client connection contains any of them, the key and the corresponding value will be removed silently during engine bootstrap and connection setup. Note that this rule is for server-side protection defined via administrators to prevent some essential configs from tampering but will not forbid users to set dynamic configurations via SET syntax. | seq | 1.2.0 | |
| kyuubi.session.conf .restrict.list | A comma separated list of restricted keys. If the client connection contains any of them, the connection will be rejected explicitly during engine bootstrap and connection setup. Note that this rule is for server-side protection defined via administrators to prevent some essential configs from tampering but will not forbid users to set dynamic configurations via SET syntax. | seq | 1.2.0 | |
| kyuubi.session.engine .check.interval | PT5M | The check interval for engine timeout | duration | 1.0.0 |
| kyuubi.session.engine .idle.timeout | PT30M | engine timeout, the engine will self-terminate when it's not accessed for this duration | duration | 1.0.0 |
| kyuubi.session.engine .initialize.timeout | PT1M | Timeout for starting the background engine, e.g. SparkSQLEngine. | duration | 1.0.0 |
| kyuubi.session.engine .log.timeout | PT24H | If we use Spark as the engine then the session submit log is the console output of spark-submit. We will retain the session submit log until over the config value. | duration | 1.1.0 |
| kyuubi.session.engine .login.timeout | PT15S | The timeout(ms) of creating the connection to remote sql query engine | duration | 1.0.0 |
| kyuubi.session.engine .share.level | USER | (deprecated) - Using kyuubi.engine.share.level instead | string | 1.0.0 |
| kyuubi.session.engine .spark.main.resource | <undefined> | The package used to create Spark SQL engine remote application. If it is undefined, Kyuubi will use the default | string | 1.0.0 |
| kyuubi.session.engine .startup.error.max .size | 8192 | During engine bootstrapping, if error occurs, using this config to limit the length error message(characters). | int | 1.1.0 |
| kyuubi.session.idle .timeout | PT6H | session idle timeout, it will be closed when it's not accessed for this duration | duration | 1.2.0 |
| kyuubi.session .timeout | PT6H | (deprecated)session timeout, it will be closed when it's not accessed for this duration | duration | 1.0.0 |
Zookeeper
| Key | Default | Meaning | Type | Since |
|---|---|---|---|---|
| kyuubi.zookeeper .embedded.client.port | 2181 | clientPort for the embedded zookeeper server to listen for client connections, a client here could be Kyuubi server, engine and JDBC client | int | 1.2.0 |
| kyuubi.zookeeper .embedded.client.port .address | <undefined> | clientPortAddress for the embedded zookeeper server to | string | 1.2.0 |
| kyuubi.zookeeper .embedded.data.dir | embedded_zookeeper | dataDir for the embedded zookeeper server where stores the in-memory database snapshots and, unless specified otherwise, the transaction log of updates to the database. | string | 1.2.0 |
| kyuubi.zookeeper .embedded.data.log.dir | embedded_zookeeper | dataLogDir for the embedded zookeeper server where writes the transaction log . | string | 1.2.0 |
| kyuubi.zookeeper .embedded.directory | embedded_zookeeper | The temporary directory for the embedded zookeeper server | string | 1.0.0 |
| kyuubi.zookeeper .embedded.election .port | 0 | electionPort for the embedded zookeeper server | int | 1.2.0 |
| kyuubi.zookeeper .embedded.max.client .connections | 120 | maxClientCnxns for the embedded zookeeper server to limits the number of concurrent connections of a single client identified by IP address | int | 1.2.0 |
| kyuubi.zookeeper .embedded.max.session .timeout | <undefined> | maxSessionTimeout in milliseconds for the embedded zookeeper server will allow the client to negotiate. Defaults to 20 times the tickTime | int | 1.2.0 |
| kyuubi.zookeeper .embedded.min.session .timeout | <undefined> | minSessionTimeout in milliseconds for the embedded zookeeper server will allow the client to negotiate. Defaults to 2 times the tickTime | int | 1.2.0 |
| kyuubi.zookeeper .embedded.port | 2181 | The port of the embedded zookeeper server | int | 1.0.0 |
| kyuubi.zookeeper .embedded.quorum.port | 0 | quorumPort for the embedded zookeeper server | int | 1.2.0 |
| kyuubi.zookeeper .embedded.server.id | -1 | serverId for the embedded zookeeper server | int | 1.2.0 |
| kyuubi.zookeeper .embedded.tick.time | 3000 | tickTime in milliseconds for the embedded zookeeper server | int | 1.2.0 |
Spark Configurations
Via spark-defaults.conf
Setting them in $SPARK_HOME/conf/spark-defaults.conf supplies with default values for SQL engine application. Available properties can be found at Spark official online documentation for Spark Configurations
Via kyuubi-defaults.conf
Setting them in $KYUUBI_HOME/conf/kyuubi-defaults.conf supplies with default values for SQL engine application too. These properties will override all settings in $SPARK_HOME/conf/spark-defaults.conf
Via JDBC Connection URL
Setting them in the JDBC Connection URL supplies session-specific for each SQL engine. For example: jdbc:hive2://localhost:10009/default;#spark.sql.shuffle.partitions=2;spark.executor.memory=5g
Runtime SQL Configuration
- For Runtime SQL Configurations, they will take affect every time
Static SQL and Spark Core Configuration
- For Static SQL Configurations and other spark core configs, e.g.
spark.executor.memory, they will take affect if there is no existing SQL engine application. Otherwise, they will just be ignored
- For Static SQL Configurations and other spark core configs, e.g.
Via SET Syntax
Please refer to the Spark official online documentation for SET Command
Logging
Kyuubi uses log4j for logging. You can configure it using $KYUUBI_HOME/conf/log4j.properties.
# # Licensed to the Apache Software Foundation (ASF) under one or more # contributor license agreements. See the NOTICE file distributed with # this work for additional information regarding copyright ownership. # The ASF licenses this file to You under the Apache License, Version 2.0 # (the "License"); you may not use this file except in compliance with # the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # Set everything to be logged to the console log4j.rootCategory=INFO, console log4j.appender.console=org.apache.log4j.ConsoleAppender log4j.appender.console.target=System.err log4j.appender.console.layout=org.apache.log4j.PatternLayout log4j.appender.console.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss.SSS} %p %c{2}: %m%n
Other Configurations
Hadoop Configurations
Specifying HADOOP_CONF_DIR to the directory contains hadoop configuration files or treating them as Spark properties with a spark.hadoop. prefix. Please refer to the Spark official online documentation for Inheriting Hadoop Cluster Configuration. Also, please refer to the Apache Hadoop's online documentation for an overview on how to configure Hadoop.
Hive Configurations
These configurations are used for SQL engine application to talk to Hive MetaStore and could be configured in a hive-site.xml. Placed it in $SPARK_HOME/conf directory, or treating them as Spark properties with a spark.hadoop. prefix.
User Defaults
In Kyuubi, we can configure user default settings to meet separate needs. These user defaults override system defaults, but will be overridden by those from JDBC Connection URL or Set Command if could be. They will take effect when creating the SQL engine application ONLY. User default settings are in the form of ___{username}___.{config key}. There are three continuous underscores(_) at both sides of the username and a dot(.) that separates the config key and the prefix. For example:
# For system defaults spark.master=local spark.sql.adaptive.enabled=true # For a user named kent ___kent___.spark.master=yarn ___kent___.spark.sql.adaptive.enabled=false # For a user named bob ___bob___.spark.master=spark://master:7077 ___bob___.spark.executor.memory=8g
In the above case, if there are related configurations from JDBC Connection URL, kent will run his SQL engine application on YARN and prefer the Spark AQE to be off, while bob will activate his SQL engine application on a Spark standalone cluster with 8g heap memory for each executor and obey the Spark AQE behavior of Kyuubi system default. On the other hand, for those users who do not have custom configurations will use system defaults.