Added: knox/trunk/books/1.3.0/config.md URL: http://svn.apache.org/viewvc/knox/trunk/books/1.3.0/config.md?rev=1850181&view=auto ============================================================================== --- knox/trunk/books/1.3.0/config.md (added) +++ knox/trunk/books/1.3.0/config.md Wed Jan 2 17:31:29 2019 @@ -0,0 +1,983 @@ +<!-- + 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. +--> + +### Configuration ### + +Configuration for Apache Knox includes: + +1. #[Related Cluster Configuration] that must be done within the Hadoop cluster to allow Knox to communicate with various services +2. #[Gateway Server Configuration] - which is the configurable elements of the server itself which applies to behavior that spans all topologies or managed Hadoop clusters +3. #[Topology Descriptors] which are the descriptors for controlling access to Hadoop clusters in various ways + +### Related Cluster Configuration ### + +The following configuration changes must be made to your cluster to allow Apache Knox to +dispatch requests to the various service components on behalf of end users. + +#### Grant Proxy privileges for Knox user in `core-site.xml` on Hadoop master nodes #### + +Update `core-site.xml` and add the following lines towards the end of the file. + +Replace `FQDN_OF_KNOX_HOST` with the fully qualified domain name of the host running the Knox gateway. +You can usually find this by running `hostname -f` on that host. + +You can use `*` for local developer testing if the Knox host does not have a static IP. + + <property> + <name>hadoop.proxyuser.knox.groups</name> + <value>users</value> + </property> + <property> + <name>hadoop.proxyuser.knox.hosts</name> + <value>FQDN_OF_KNOX_HOST</value> + </property> + +#### Grant proxy privilege for Knox in `webhcat-site.xml` on Hadoop master nodes #### + +Update `webhcat-site.xml` and add the following lines towards the end of the file. + +Replace `FQDN_OF_KNOX_HOST` with the fully qualified domain name of the host running the Knox gateway. +You can use `*` for local developer testing if the Knox host does not have a static IP. + + <property> + <name>webhcat.proxyuser.knox.groups</name> + <value>users</value> + </property> + <property> + <name>webhcat.proxyuser.knox.hosts</name> + <value>FQDN_OF_KNOX_HOST</value> + </property> + +#### Grant proxy privilege for Knox in `oozie-site.xml` on Oozie host #### + +Update `oozie-site.xml` and add the following lines towards the end of the file. + +Replace `FQDN_OF_KNOX_HOST` with the fully qualified domain name of the host running the Knox gateway. +You can use `*` for local developer testing if the Knox host does not have a static IP. + + <property> + <name>oozie.service.ProxyUserService.proxyuser.knox.groups</name> + <value>users</value> + </property> + <property> + <name>oozie.service.ProxyUserService.proxyuser.knox.hosts</name> + <value>FQDN_OF_KNOX_HOST</value> + </property> + +#### Enable http transport mode and use substitution in HiveServer2 #### + +Update `hive-site.xml` and set the following properties on HiveServer2 hosts. +Some of the properties may already be in the hive-site.xml. +Ensure that the values match the ones below. + + <property> + <name>hive.server2.allow.user.substitution</name> + <value>true</value> + </property> + + <property> + <name>hive.server2.transport.mode</name> + <value>http</value> + <description>Server transport mode. "binary" or "http".</description> + </property> + + <property> + <name>hive.server2.thrift.http.port</name> + <value>10001</value> + <description>Port number when in HTTP mode.</description> + </property> + + <property> + <name>hive.server2.thrift.http.path</name> + <value>cliservice</value> + <description>Path component of URL endpoint when in HTTP mode.</description> + </property> + +#### Gateway Server Configuration #### + +The following table illustrates the configurable elements of the Apache Knox Gateway at the server level via gateway-site.xml. + +Property | Description | Default +------------|-----------|----------- +`gateway.deployment.dir`|The directory within `GATEWAY_HOME` that contains gateway topology deployments|`{GATEWAY_HOME}/data/deployments` +`gateway.security.dir`|The directory within `GATEWAY_HOME` that contains the required security artifacts|`{GATEWAY_HOME}/data/security` +`gateway.data.dir`|The directory within `GATEWAY_HOME` that contains the gateway instance data|`{GATEWAY_HOME}/data` +`gateway.services.dir`|The directory within `GATEWAY_HOME` that contains the gateway services definitions|`{GATEWAY_HOME}/services` +`gateway.hadoop.conf.dir`|The directory within `GATEWAY_HOME` that contains the gateway configuration|`{GATEWAY_HOME}/conf` +`gateway.frontend.url`|The URL that should be used during rewriting so that it can rewrite the URLs with the correct "frontend" URL|none +`gateway.xforwarded.enabled`|Indicates whether support for some X-Forwarded-* headers is enabled|`true` +`gateway.trust.all.certs`|Indicates whether all presented client certs should establish trust|`false` +`gateway.client.auth.needed`|Indicates whether clients are required to establish a trust relationship with client certificates|`false` +`gateway.truststore.path`|Location of the truststore for client certificates to be trusted|`gateway.jks` +`gateway.truststore.type`|Indicates the type of truststore|`JKS` +`gateway.keystore.type`|Indicates the type of keystore for the identity store|`JKS` +`gateway.jdk.tls.ephemeralDHKeySize`|`jdk.tls.ephemeralDHKeySize`, is defined to customize the ephemeral DH key sizes. The minimum acceptable DH key size is 1024 bits, except for exportable cipher suites or legacy mode (`jdk.tls.ephemeralDHKeySize=legacy`)|`2048` +`gateway.threadpool.max`|The maximum concurrent requests the server will process. The default is 254. Connections beyond this will be queued.|`254` +`gateway.httpclient.maxConnections`|The maximum number of connections that a single HttpClient will maintain to a single host:port.|`32` +`gateway.httpclient.connectionTimeout`|The amount of time to wait when attempting a connection. The natural unit is milliseconds, but a 's' or 'm' suffix may be used for seconds or minutes respectively.|20s +`gateway.httpclient.socketTimeout`|The amount of time to wait for data on a socket before aborting the connection. The natural unit is milliseconds, but a 's' or 'm' suffix may be used for seconds or minutes respectively.| 20s +`gateway.httpserver.requestBuffer`|The size of the HTTP server request buffer in bytes|`16384` +`gateway.httpserver.requestHeaderBuffer`|The size of the HTTP server request header buffer in bytes|`8192` +`gateway.httpserver.responseBuffer`|The size of the HTTP server response buffer in bytes|`32768` +`gateway.httpserver.responseHeaderBuffer`|The size of the HTTP server response header buffer in bytes|`8192` +`gateway.websocket.feature.enabled`|Enable/Disable WebSocket feature|`false` +`gateway.signing.keystore.name`|OPTIONAL Filename of keystore file that contains the signing keypair. NOTE: An alias needs to be created using `knoxcli.sh create-alias` for the alias name `signing.key.passphrase` in order to provide the passphrase to access the keystore.|null +`gateway.signing.key.alias`|OPTIONAL alias for the signing keypair within the keystore specified via `gateway.signing.keystore.name`|null +`ssl.enabled`|Indicates whether SSL is enabled for the Gateway|`true` +`ssl.include.ciphers`|A comma or pipe separated list of ciphers to accept for SSL. See the [JSSE Provider docs](http://docs.oracle.com/javase/8/docs/technotes/guides/security/SunProviders.html#SunJSSEProvider) for possible ciphers. These can also contain regular expressions as shown in the [Jetty documentation](http://www.eclipse.org/jetty/documentation/current/configuring-ssl.html).|all +`ssl.exclude.ciphers`|A comma or pipe separated list of ciphers to reject for SSL. See the [JSSE Provider docs](http://docs.oracle.com/javase/8/docs/technotes/guides/security/SunProviders.html#SunJSSEProvider) for possible ciphers. These can also contain regular expressions as shown in the [Jetty documentation](http://www.eclipse.org/jetty/documentation/current/configuring-ssl.html).|none +`ssl.exclude.protocols`|Excludes a comma or pipe separated list of protocols to not accept for SSL or "none"|`SSLv3` +`gateway.remote.config.monitor.client`|A reference to the [remote configuration registry client](#Remote+Configuration+Registry+Clients) the remote configuration monitor will employ|null +`gateway.remote.config.monitor.client.allowUnauthenticatedReadAccess` | When a remote registry client is configured to access a registry securely, this property can be set to allow unauthenticated clients to continue to read the content from that registry by setting the ACLs accordingly. | `false` +`gateway.remote.config.registry.<name>`|A named [remote configuration registry client](#Remote+Configuration+Registry+Clients) definition, where _name_ is an arbitrary identifier for the connection|null +`gateway.cluster.config.monitor.ambari.enabled`| Indicates whether the cluster monitoring and associated dynamic topology updating is enabled | `false` +`gateway.cluster.config.monitor.ambari.interval` | The interval (in seconds) at which the cluster monitor will poll Ambari for cluster configuration changes | `60` +`gateway.remote.alias.service.enabled` | Turn on/off remote alias management, this will take effect only when remote configuration monitoring is enabled | `true` +`gateway.read.only.override.topologies` | A comma-delimited list of topology names which should be forcibly treated as read-only. | none +`gateway.discovery.default.address` | The default discovery address, which is applied if no address is specified in a descriptor. | null +`gateway.discovery.default.cluster` | The default discovery cluster name, which is applied if no cluster name is specified in a descriptor. | null +`gateway.dispatch.whitelist` | A semicolon-delimited list of regular expressions for controlling to which endpoints Knox dispatches and redirects will be permitted. If DEFAULT is specified, or the property is omitted entirely, then a default domain-based whitelist will be derived from the Knox host. An empty value means no dispatches will be permitted. | null +`gateway.dispatch.whitelist.services` | A comma-delimited list of service roles to which the *gateway.dispatch.whitelist* will be applied. | none +`gateway.strict.topology.validation` | If true, topology XML files will be validated against the topology schema during redeploy | `false` +`gateway.global.rules.services` | Set the list of service names that have global rules, all services that are not in this list have rules that are treated as scoped to only to that service. | `"NAMENODE","JOBTRACKER", "WEBHDFS", "WEBHCAT", "OOZIE", "WEBHBASE", "HIVE", "RESOURCEMANAGER"` + +#### Topology Descriptors #### + +The topology descriptor files provide the gateway with per-cluster configuration information. +This includes configuration for both the providers within the gateway and the services within the Hadoop cluster. +These files are located in `{GATEWAY_HOME}/conf/topologies`. +The general outline of this document looks like this. + + <topology> + <gateway> + <provider> + </provider> + </gateway> + <service> + </service> + </topology> + +There are typically multiple `<provider>` and `<service>` elements. + +/topology +: Defines the provider and configuration and service topology for a single Hadoop cluster. + +/topology/gateway +: Groups all of the provider elements + +/topology/gateway/provider +: Defines the configuration of a specific provider for the cluster. + +/topology/service +: Defines the location of a specific Hadoop service within the Hadoop cluster. + +##### Provider Configuration ##### + +Provider configuration is used to customize the behavior of a particular gateway feature. +The general outline of a provider element looks like this. + + <provider> + <role>authentication</role> + <name>ShiroProvider</name> + <enabled>true</enabled> + <param> + <name></name> + <value></value> + </param> + </provider> + +/topology/gateway/provider +: Groups information for a specific provider. + +/topology/gateway/provider/role +: Defines the role of a particular provider. +There are a number of pre-defined roles used by out-of-the-box provider plugins for the gateway. +These roles are: authentication, identity-assertion, rewrite and hostmap + +/topology/gateway/provider/name +: Defines the name of the provider for which this configuration applies. +There can be multiple provider implementations for a given role. +Specifying the name is used to identify which particular provider is being configured. +Typically each topology descriptor should contain only one provider for each role but there are exceptions. + +/topology/gateway/provider/enabled +: Allows a particular provider to be enabled or disabled via `true` or `false` respectively. +When a provider is disabled any filters associated with that provider are excluded from the processing chain. + +/topology/gateway/provider/param +: These elements are used to supply provider configuration. +There can be zero or more of these per provider. + +/topology/gateway/provider/param/name +: The name of a parameter to pass to the provider. + +/topology/gateway/provider/param/value +: The value of a parameter to pass to the provider. + +##### Service Configuration ##### + +Service configuration is used to specify the location of services within the Hadoop cluster. +The general outline of a service element looks like this. + + <service> + <role>WEBHDFS</role> + <url>http://localhost:50070/webhdfs</url> + </service> + +/topology/service +: Provider information about a particular service within the Hadoop cluster. +Not all services are necessarily exposed as gateway endpoints. + +/topology/service/role +: Identifies the role of this service. +Currently supported roles are: WEBHDFS, WEBHCAT, WEBHBASE, OOZIE, HIVE, NAMENODE, JOBTRACKER, RESOURCEMANAGER +Additional service roles can be supported via plugins. Note: The role names are case sensitive and must be upper case. + +topology/service/url +: The URL identifying the location of a particular service within the Hadoop cluster. + +#### Hostmap Provider #### + +The purpose of the Hostmap provider is to handle situations where hosts are known by one name within the cluster and another name externally. +This frequently occurs when virtual machines are used and in particular when using cloud hosting services. +Currently, the Hostmap provider is configured as part of the topology file. +The basic structure is shown below. + + <topology> + <gateway> + ... + <provider> + <role>hostmap</role> + <name>static</name> + <enabled>true</enabled> + <param><name>external-host-name</name><value>internal-host-name</value></param> + </provider> + ... + </gateway> + ... + </topology> + +This mapping is required because the Hadoop services running within the cluster are unaware that they are being accessed from outside the cluster. +Therefore URLs returned as part of REST API responses will typically contain internal host names. +Since clients outside the cluster will be unable to resolve those host name they must be mapped to external host names. + +##### Hostmap Provider Example - EC2 ##### + +Consider an EC2 example where two VMs have been allocated. +Each VM has an external host name by which it can be accessed via the internet. +However the EC2 VM is unaware of this external host name and instead is configured with the internal host name. + + External HOSTNAMES: + ec2-23-22-31-165.compute-1.amazonaws.com + ec2-23-23-25-10.compute-1.amazonaws.com + + Internal HOSTNAMES: + ip-10-118-99-172.ec2.internal + ip-10-39-107-209.ec2.internal + +The Hostmap configuration required to allow access external to the Hadoop cluster via the Apache Knox Gateway would be this: + + <topology> + <gateway> + ... + <provider> + <role>hostmap</role> + <name>static</name> + <enabled>true</enabled> + <param> + <name>ec2-23-22-31-165.compute-1.amazonaws.com</name> + <value>ip-10-118-99-172.ec2.internal</value> + </param> + <param> + <name>ec2-23-23-25-10.compute-1.amazonaws.com</name> + <value>ip-10-39-107-209.ec2.internal</value> + </param> + </provider> + ... + </gateway> + ... + </topology> + +##### Hostmap Provider Example - Sandbox ##### + +The Hortonworks Sandbox 2.x poses a different challenge for host name mapping. +This version of the Sandbox uses port mapping to make the Sandbox VM appear as though it is accessible via localhost. +However the Sandbox VM is internally configured to consider sandbox.hortonworks.com as the host name. +So from the perspective of a client accessing Sandbox the external host name is localhost. +The Hostmap configuration required to allow access to Sandbox from the host operating system is this. + + <topology> + <gateway> + ... + <provider> + <role>hostmap</role> + <name>static</name> + <enabled>true</enabled> + <param> + <name>localhost</name> + <value>sandbox,sandbox.hortonworks.com</value> + </param> + </provider> + ... + </gateway> + ... + </topology> + +##### Hostmap Provider Configuration ##### + +Details about each provider configuration element is enumerated below. + +topology/gateway/provider/role +: The role for a Hostmap provider must always be `hostmap`. + +topology/gateway/provider/name +: The Hostmap provider supplied out-of-the-box is selected via the name `static`. + +topology/gateway/provider/enabled +: Host mapping can be enabled or disabled by providing `true` or `false`. + +topology/gateway/provider/param +: Host mapping is configured by providing parameters for each external to internal mapping. + +topology/gateway/provider/param/name +: The parameter names represent the external host names associated with the internal host names provided by the value element. +This can be a comma separated list of host names that all represent the same physical host. +When mapping from internal to external host name the first external host name in the list is used. + +topology/gateway/provider/param/value +: The parameter values represent the internal host names associated with the external host names provider by the name element. +This can be a comma separated list of host names that all represent the same physical host. +When mapping from external to internal host names the first internal host name in the list is used. + + +#### Simplified Topology Descriptors #### + +Simplified descriptors are a means to facilitate provider configuration sharing and service endpoint discovery. Rather than +editing an XML topology descriptor, it's possible to create a simpler YAML (or JSON) descriptor specifying the desired contents +of a topology, which will yield a full topology descriptor and deployment. + +##### Externalized Provider Configurations ##### + +Sometimes, the same provider configuration is applied to multiple Knox topologies. With the provider configuration externalized +from the simple descriptors, a single configuration can be referenced by multiple topologies. This helps reduce the duplication +of configuration, and the need to update multiple configuration files when a policy change is required. Updating a provider +configuration will trigger an update to all those topologies that reference it. + +The contents of externalized provider configuration details are identical to the contents of the gateway element from a full topology descriptor. +The only difference is that those details are defined in a separate JSON/YAML file in `{GATEWAY_HOME}/conf/shared-providers/`, which is then +referenced by one or more descriptors. + +*Provider Configuration Example* + + { + "providers": [ + { + "role": "authentication", + "name": "ShiroProvider", + "enabled": "true", + "params": { + "sessionTimeout": "30", + "main.ldapRealm": "org.apache.knox.gateway.shirorealm.KnoxLdapRealm", + "main.ldapContextFactory": "org.apache.knox.gateway.shirorealm.KnoxLdapContextFactory", + "main.ldapRealm.contextFactory": "$ldapContextFactory", + "main.ldapRealm.userDnTemplate": "uid={0},ou=people,dc=hadoop,dc=apache,dc=org", + "main.ldapRealm.contextFactory.url": "ldap://localhost:33389";, + "main.ldapRealm.contextFactory.authenticationMechanism": "simple", + "urls./**": "authcBasic" + } + }, + { + "name": "static", + "role": "hostmape", + "enabled": "true", + "params": { + "localhost": "sandbox,sandbox.hortonworks.com" + } + } + ] + } + + + +###### Sharing HA Providers ###### + +HA Providers are a special concern with respect to sharing provider configuration because they include service-specific (and possibly cluster-specific) configuration. + +This requires extra attention because the service configurations corresponding to the associated HA Provider configuration must contain the correct content to function properly. + +For a shared provider configuration with an HA Provider service: + +- If the referencing descriptor does not declare the corresponding service, then the HA Provider configuration is effectively ignored since the service isn't exposed by the topology. +- If a corresponding service is declared in the descriptor + - If service endpoint discovery is employed, then Knox should populate the URLs correctly to support the HA behavior. + - Otherwise, the URLs must be explicitly specified for that service in the descriptor. +- If the descriptor content is correct, but the cluster service is not configured for HA, then the HA behavior obviously won't work. + +__*Apache ZooKeeper-based HA Provider Services*__ + +The HA Provider configuration for some services (e.g., [HiveServer2](#HiveServer2+HA), [Kafka](#Kafka+HA)) includes references to Apache ZooKeeper hosts (i.e., the ZooKeeper ensemble) +and namespaces. It's important to understand the relationship of that ensemble configuration to the topologies referencing it. These ZooKeeper details are often cluster-specific. If the ZooKeeper ensemble in the provider configuration is part of cluster *A*, then it's probably incorrect to reference it in a topology for cluster *B* since the Hadoop service endpoints will probably be the wrong ones. However, if multiple clusters are working with a common ZooKeeper ensemble, then sharing this provider configuration *may* be appropriate. + +*It's always best to specify cluster-specific details in a descriptor rather than a provider configuration.* + +All of the service attributes, which can be specified in the HaProvider, can also be specified as params in the corresponding service declaration in the descriptor. If an attribute is specified in both the service declaration and the HaProvider, then the service-level value __overrides__ the HaProvider-level value. + + "services": [ + { + "name": "HIVE", + "params": { + "enabled": "true", + "zookeeperEnsemble": "host1:2181,host2:2181,host3:2181", + "zookeeperNamespace" : "hiveserver2", + "maxRetryAttempts" : "100" + } + } + ] + +Note that Knox can dynamically determine these ZooKeeper ensemble details for *some* services; for others, they are static provider configuration details. +The services for which Knox can discover the cluster-specific ZooKeeper details include: + +- YARN +- HIVE +- WEBHDFS +- WEBHBASE +- WEBHCAT +- OOZIE +- ATLAS +- ATLAS-API +- KAFKA + +For a subset of these supported services, Knox can also determine whether ZooKeeper-based HA is enabled or not. +This means that the *enabled* attribute of the HA Provider configuration for these services may be set to __auto__, and Knox will determine whether or not it is enabled based on that service's configuration in the target cluster. + + { + "providers": [ + { + "role": "ha", + "name": "HaProvider", + "enabled": "true", + "params": { + "WEBHDFS": "maxFailoverAttempts=3;failoverSleep=1000;maxRetryAttempts=3;retrySleep=1000;enabled=true", + "HIVE": "maxFailoverAttempts=10;failoverSleep=1000;maxRetryAttempts=5;retrySleep=1000;enabled=auto", + "YARN": "maxFailoverAttempts=5;failoverSleep=5000;maxRetryAttempts=3;retrySleep=1000;enabled=auto" + } + } + ] + } + +These services include: + +- YARN +- HIVE +- ATLAS +- ATLAS-API + + +Be sure to pay extra attention when sharing HA Provider configuration across topologies. + + +##### Simplified Descriptor Files ##### + +Simplified descriptors allow service URLs to be defined explicitly, just like full topology descriptors. However, if URLs are +omitted for a service, Knox will attempt to discover that service's URLs from the Hadoop cluster. Currently, this behavior is +only supported for clusters managed by Ambari. In any case, the simplified descriptors are much more concise than a full +topology descriptor. + +*Descriptor Properties* + +Property | Description +------------|----------- +`discovery-type`|The discovery source type. (Currently, the only supported type is `AMBARI`). +`discovery-address`|The endpoint address for the discovery source. +`discovery-user`|The username with permission to access the discovery source. If omitted, then Knox will check for an alias named `ambari.discovery.user`, and use its value if defined. +`discovery-pwd-alias`|The alias of the password for the user with permission to access the discovery source. If omitted, then Knox will check for an alias named `ambari.discovery.password`, and use its value if defined. +`provider-config-ref`|A reference to a provider configuration in `{GATEWAY_HOME}/conf/shared-providers/`. +`cluster`|The name of the cluster from which the topology service endpoints should be determined. +`services`|The collection of services to be included in the topology. +`applications`|The collection of applications to be included in the topology. + + +Two file formats are supported for two distinct purposes. + +- YAML is intended for the individual hand-editing a simplified descriptor because of its readability. +- JSON is intended to be used for [API](#Admin+API) interactions. + +That being said, there is nothing preventing the hand-editing of files in the JSON format. However, the API will *not* accept YAML files as input. + +*YAML Example* (based on the HDP Docker Sandbox) + + --- + # Discovery source config + discovery-type : AMBARI + discovery-address : http://sandbox.hortonworks.com:8080 + + # If this is not specified, the alias ambari.discovery.user is checked for a username + discovery-user : maria_dev + + # If this is not specified, the default alias ambari.discovery.password is used + discovery-pwd-alias : sandbox.discovery.password + + # Provider config reference, the contents of which will be included in the resulting topology descriptor + provider-config-ref : sandbox-providers + + # The cluster for which the details should be discovered + cluster: Sandbox + + # The services to declare in the resulting topology descriptor, whose URLs will be discovered (unless a value is specified) + services: + - name: NAMENODE + - name: JOBTRACKER + - name: WEBHDFS + - name: WEBHCAT + - name: OOZIE + - name: WEBHBASE + - name: HIVE + - name: RESOURCEMANAGER + - name: KNOXSSO + params: + knoxsso.cookie.secure.only: true + knoxsso.token.ttl: 100000 + - name: AMBARI + urls: + - http://sandbox.hortonworks.com:8080 + - name: AMBARIUI + urls: + - http://sandbox.hortonworks.com:8080 + - name: AMBARIWS + urls: + - ws://sandbox.hortonworks.com:8080 + + +*JSON Example* (based on the HDP Docker Sandbox) + + { + "discovery-type":"AMBARI", + "discovery-address":"http://sandbox.hortonworks.com:8080";, + "discovery-user":"maria_dev", + "discovery-pwd-alias":"sandbox.discovery.password", + "provider-config-ref":"sandbox-providers", + "cluster":"Sandbox", + "services":[ + {"name":"NAMENODE"}, + {"name":"JOBTRACKER"}, + {"name":"WEBHDFS"}, + {"name":"WEBHCAT"}, + {"name":"OOZIE"}, + {"name":"WEBHBASE"}, + {"name":"HIVE"}, + {"name":"RESOURCEMANAGER"}, + {"name":"KNOXSSO", + "params":{ + "knoxsso.cookie.secure.only":"true", + "knoxsso.token.ttl":"100000" + } + }, + {"name":"AMBARI", "urls":["http://sandbox.hortonworks.com:8080"]}, + {"name":"AMBARIUI", "urls":["http://sandbox.hortonworks.com:8080";], + {"name":"AMBARIWS", "urls":["ws://sandbox.hortonworks.com:8080"]} + ] + } + + +Both of these examples illustrate the specification of credentials for the interaction with Ambari. If no credentials are specified, then the default +aliases are queried. Use of the default aliases is sufficient for scenarios where topology discovery will only interact with a single Ambari instance. +For multiple Ambari instances however, it's most likely that each will require different sets of credentials. The discovery-user and discovery-pwd-alias +properties exist for this purpose. Note that whether using the default credential aliases or specifying a custom password alias, these +[aliases must be defined](#Alias+creation) prior to any attempt to deploy a topology using a simplified descriptor. + + +##### Deployment Directories ##### + +Effecting topology changes is as simple as modifying files in two specific directories. + +The `{GATEWAY_HOME}/conf/shared-providers/` directory is the location where Knox looks for provider configurations. This directory is monitored for changes, such +that modifying a provider configuration file therein will trigger updates to any referencing simplified descriptors in the `{GATEWAY_HOME}/conf/descriptors/` directory. +*Care should be taken when deleting these files if there are referencing descriptors; any subsequent modifications of referencing descriptors will fail when the deleted +provider configuration cannot be found. The references should all be modified before deleting the provider configuration.* + +Likewise, the `{GATEWAY_HOME}/conf/descriptors/` directory is monitored for changes, such that adding or modifying a simplified descriptor file in this directory will +trigger the generation and deployment of a topology descriptor. Deleting a descriptor from this directory will conversely result in the removal of the previously-generated +topology descriptor, and the associated topology will be undeployed. + +If the service details for a deployed (generated) topology are changed in the cluster, then the Knox topology can be updated by 'touch'ing the simplified descriptor. This +will trigger discovery and regeneration/redeployment of the topology descriptor. + +Note that deleting a generated topology descriptor from `{GATEWAY_HOME}/conf/topologies/` is not sufficient for its removal. If the source descriptor is modified, or Knox +is restarted, the topology descriptor will be regenerated and deployed. Removing generated topology descriptors should be done by removing the associated simplified descriptor. +For the same reason, editing generated topology descriptors is strongly discouraged since they can be inadvertently overwritten. + + +Another means by which these topology changes can be effected is the [Admin API](#Admin+API). + +##### Cloud Federation Configuration ##### + +Cloud Federation feature allows for a topology based federation from one Knox instance to another (from on-prem Knox instance to cloud knox instance). + +##### Cluster Configuration Monitoring ##### + +Another benefit gained through the use of simplified topology descriptors, and the associated service discovery, is the ability to monitor clusters for configuration changes. +__Like service discovery, this is currently only available for clusters managed by Ambari.__ + +The gateway can monitor Ambari cluster configurations, and respond to changes by dynamically regenerating and redeploying the affected topologies. +The following properties in gateway-site.xml can be used to control this behavior. + + <property> + <name>gateway.cluster.config.monitor.ambari.enabled</name> + <value>false</value> + <description>Enable/disable Ambari cluster configuration monitoring.</description> + </property> + + <property> + <name>gateway.cluster.config.monitor.ambari.interval</name> + <value>60</value> + <description>The interval (in seconds) for polling Ambari for cluster configuration changes.</description> + </property> + + +Since service discovery supports multiple Ambari instances as discovery sources, multiple Ambari instances can be monitored for cluster configuration changes. + +For example, if the cluster monitor is enabled, deployment of the following simple descriptor would trigger monitoring of the *Sandbox* cluster managed by Ambari @ http://sandbox.hortonworks.com:8080 + + --- + discovery-address : http://sandbox.hortonworks.com:8080 + discovery-user : maria_dev + discovery-pwd-alias : sandbox.discovery.password + cluster: Sandbox + provider-config-ref : sandbox-providers + services: + - name: NAMENODE + - name: JOBTRACKER + - name: WEBHDFS + - name: WEBHCAT + - name: OOZIE + - name: WEBHBASE + - name: HIVE + - name: RESOURCEMANAGER + +Another *Sandbox* cluster, managed by a __different__ Ambari instance, could simultaneously be monitored by the same gateway instance. + +Now, topologies can be kept in sync with their respective target cluster configurations, without administrator intervention or service interruption. + + + +##### Remote Configuration Monitor ##### + +In addition to monitoring local directories for provider configurations and simplified descriptors, the gateway similarly supports monitoring ZooKeeper. + +This monitor depends on a [remote configuration registry client](#Remote+Configuration+Registry+Clients), and that client must be specified by setting the following property in gateway-site.xml + + <property> + <name>gateway.remote.config.monitor.client</name> + <value>sandbox-zookeeper-client</value> + <description>Remote configuration monitor client name.</description> + </property> + +This client identifier is a reference to a remote configuration registry client, as in this example (also defined in gateway-site.xml) + + <property> + <name>gateway.remote.config.registry.sandbox-zookeeper-client</name> + <value>type=ZooKeeper;address=localhost:2181</value> + <description>ZooKeeper configuration registry client details.</description> + </property> + +_The actual name of the client (e.g., sandbox-zookeeper-client) is not important, except that the reference matches the name specified in the client definition._ + +With this configuration, the gateway will monitor the following znodes in the specified ZooKeeper instance + + /knox + /config + /shared-providers + /descriptors + +The creation of these znodes, and the population of their respective contents, is an activity __not__ currently managed by the gateway. However, the [KNOX CLI](#Knox+CLI) includes commands for managing the contents +of these znodes. + +These znodes are treated similarly to the local *shared-providers* and *descriptors* directories described in [Deployment Directories](#Deployment+Directories). +When the monitor notices a change to these znodes, it will attempt to effect the same change locally. + +If a provider configuration is added to the */knox/config/shared-providers* znode, the monitor will download the new configuration to the local shared-providers directory. +Likewise, if a descriptor is added to the */knox/config/descriptors* znode, the monitor will download the new descriptor to the local descriptors directory, which will +trigger an attempt to generate and deploy a corresponding topology. + +Modifications to the contents of these znodes, will yield the same behavior as can be seen resulting from the corresponding local modification. + +znode | action | result +------|--------|-------- +`/knox/config/shared-providers` | add | Download the new file to the local shared-providers directory +`/knox/config/shared-providers` | modify | Download the new file to the local shared-providers directory; If there are any existing descriptor references, then topology will be regenerated and redeployed for those referencing descriptors. +`/knox/config/shared-providers` | delete | Delete the corresponding file from the local shared-providers directory +`/knox/config/descriptors` | add | Download the new file to the local descriptors directory; A corresponding topology will be generated and deployed. +`/knox/config/descriptors` | modify | Download the new file to the local descriptors directory; The corresponding topology will be regenerated and redeployed. +`/knox/config/descriptors` | delete | Delete the corresponding file from the local descriptors directory + +This simplifies the configuration for HA gateway deployments, in that the gateway instances can all be configured to monitor the same ZooKeeper instance, and changes to the znodes' contents will be applied to all those gateway instances. With this approach, it is no longer necessary to manually deploy topologies to each of the gateway instances. + +_A Note About ACLs_ + + While the gateway does not currently require secure interactions with remote registries, it is recommended + that ACLs be applied to restrict at least writing of the entries referenced by this monitor. If write + access is available to everyone, then the contents of the configuration cannot be known to be trustworthy, + and there is the potential for malicious activity. Be sure to carefully consider who will have the ability + to define configuration in monitored remote registries and apply the necessary measures to ensure its + trustworthiness. + + +#### Remote Configuration Registry Clients #### + +One or more features of the gateway employ remote configuration registry (e.g., ZooKeeper) clients. These clients are configured by setting properties in the gateway configuration (`gateway-site.xml`). + +Each client configuration is a single property, the name of which is prefixed with __gateway.remote.config.registry.__ and suffixed by the client identifier. +The value of such a property, is a registry-type-specific set of semicolon-delimited properties for that client, including the type of registry with which it will interact. + + <property> + <name>gateway.remote.config.registry.a-zookeeper-client</name> + <value>type=ZooKeeper;address=zkhost1:2181,zkhost2:2181,zkhost3:2181</value> + <description>ZooKeeper configuration registry client details.</description> + </property> + +In the preceeding example, the client identifier is __a-zookeeper-client__, by way of the property name __gateway.remote.config.registry.a-zookeeper-client__. + +The property value specifies that the client is intended to interact with ZooKeeper. It also specifies the particular ZooKeeper ensemble with which it will interact; this could be a single ZooKeeper instance as well. + +The property value may also include an optional namespace, to which the client will be restricted (i.e., "chroot" the client). + + <property> + <name>gateway.remote.config.registry.a-zookeeper-client</name> + <value>type=ZooKeeper;address=zkhost1:2181,zkhost2:2181,zkhost3:2181;namespace=/knox/config</value> + <description>ZooKeeper configuration registry client details.</description> + </property> + +At least for the ZooKeeper type, authentication details may also be specified as part of the property value, for interacting with instances for which authentication is required. + +__Digest Authentication Example__ + + <property> + <name>gateway.remote.config.registry.a-zookeeper-client</name> + <value>type=ZooKeeper;address=zkhost1:2181,zkhost2:2181,zkhost3:2181;authType=Digest;principal=myzkuser;credentialAlias=myzkpass</value> + <description>ZooKeeper configuration registry client details.</description> + </property> + + +__Kerberos Authentication Example__ + + <property> + <name>gateway.remote.config.registry.a-zookeeper-client</name> + <value>type=ZooKeeper;address=zkhost1:2181,zkhost2:2181,zkhost3:2181;authType=Kerberos;principal=myzkuser;keytab=/home/user/myzk.keytab;useKeyTab=true;useTicketCache=false</value> + <description>ZooKeeper configuration registry client details.</description> + </property> + + +_While multiple such clients can be configured, for ZooKeeper clients, there is currently a limitation with respect to authentication. Multiple clients cannot each have distinct authentication configurations. This limitation is imposed by the underlying ZooKeeper client. Therefore, the clients must all be insecure (no authentication configured), or they must all authenticate to the same ZooKeeper using the same credentials._ + +The [remote configuration monitor](#Remote+Configuration+Monitor) facility uses these client configurations to perform its function. + +#### Remote Alias Discovery #### + +Knox will also monitor for remote aliases that are added, deleted or updated. By default this is turned on (if Remote Configuration Monitor is on) and will sync all the aliases. In case one wants to turn off this feature they can do so by using the property "gateway.remote.alias.service.enabled" in gateway-site.xml. Knox needs to be restarted for this change to take effect. + + <property> + <name>gateway.remote.alias.service.enabled</name> + <value>false</value> + <description>Turn on/off Remote Alias Discovery(true by default)</description> + </property> + + +#### Logging #### + +If necessary you can enable additional logging by editing the `log4j.properties` file in the `conf` directory. +Changing the `rootLogger` value from `ERROR` to `DEBUG` will generate a large amount of debug logging. +A number of useful, more fine loggers are also provided in the file. + + +#### Java VM Options #### + +TODO - Java VM options doc. + + +#### Persisting the Master Secret #### + +The master secret is required to start the server. +This secret is used to access secured artifacts by the gateway instance. +Keystore, trust stores and credential stores are all protected with the master secret. + +You may persist the master secret by supplying the *\-persist-master* switch at startup. +This will result in a warning indicating that persisting the secret is less secure than providing it at startup. +We do make some provisions in order to protect the persisted password. + +It is encrypted with AES 128 bit encryption and where possible the file permissions are set to only be accessible by the user that the gateway is running as. + +After persisting the secret, ensure that the file at `data/security/master` has the appropriate permissions set for your environment. +This is probably the most important layer of defense for master secret. +Do not assume that the encryption is sufficient protection. + +A specific user should be created to run the gateway. This user will be the only user with permissions for the persisted master file. + +See the Knox CLI section for descriptions of the command line utilities related to the master secret. + +#### Management of Security Artifacts #### + +There are a number of artifacts that are used by the gateway in ensuring the security of wire level communications, access to protected resources and the encryption of sensitive data. +These artifacts can be managed from outside of the gateway instances or generated and populated by the gateway instance itself. + +The following is a description of how this is coordinated with both standalone (development, demo, etc.) gateway instances and instances as part of a cluster of gateways in mind. + +Upon start of the gateway server we: + +1. Look for an identity store at `data/security/keystores/gateway.jks`. + The identity store contains the certificate and private key used to represent the identity of the server for SSL connections and signature creation. + * If there is no identity store we create one and generate a self-signed certificate for use in standalone/demo mode. + The certificate is stored with an alias of gateway-identity. + * If there is an identity store found than we ensure that it can be loaded using the provided master secret and that there is an alias called gateway-identity. +2. Look for a credential store at `data/security/keystores/__gateway-credentials.jceks`. + This credential store is used to store secrets/passwords that are used by the gateway. + For instance, this is where the passphrase for accessing the gateway-identity certificate is kept. + * If there is no credential store found then we create one and populate it with a generated passphrase for the alias `gateway-identity-passphrase`. + This is coordinated with the population of the self-signed cert into the identity-store. + * If a credential store is found then we ensure that it can be loaded using the provided master secret and that the expected aliases have been populated with secrets. + +Upon deployment of a Hadoop cluster topology within the gateway we: + +1. Look for a credential store for the topology. For instance, we have a sample topology that gets deployed out of the box. We look for `data/security/keystores/sandbox-credentials.jceks`. This topology specific credential store is used for storing secrets/passwords that are used for encrypting sensitive data with topology specific keys. + * If no credential store is found for the topology being deployed then one is created for it. + Population of the aliases is delegated to the configured providers within the system that will require the use of a secret for a particular task. + They may programmatically set the value of the secret or choose to have the value for the specified alias generated through the AliasService. + * If a credential store is found then we ensure that it can be loaded with the provided master secret and the configured providers have the opportunity to ensure that the aliases are populated and if not to populate them. + +By leveraging the algorithm described above we can provide a window of opportunity for management of these artifacts in a number of ways. + +1. Using a single gateway instance as a master instance the artifacts can be generated or placed into the expected location and then replicated across all of the slave instances before startup. +2. Using an NFS mount as a central location for the artifacts would provide a single source of truth without the need to replicate them over the network. Of course, NFS mounts have their own challenges. +3. Using the KnoxCLI to create and manage the security artifacts. + +See the Knox CLI section for descriptions of the command line utilities related to the security artifact management. + +#### Keystores #### +In order to provide your own certificate for use by the gateway, you will need to either import an existing key pair into a Java keystore or generate a self-signed cert using the Java keytool. + +##### Importing a key pair into a Java keystore ##### +One way to accomplish this is to start with a PKCS12 store for your key pair and then convert it to a Java keystore or JKS. + +The following example uses OpenSSL to create a PKCS12 encoded store from your provided certificate and private key that are in PEM format. + + openssl pkcs12 -export -in cert.pem -inkey key.pem > server.p12 + +The next example converts the PKCS12 store into a Java keystore (JKS). It should prompt you for the keystore and key passwords for the destination keystore. You must use the master-secret for the keystore password and keep track of the password that you use for the key passphrase. + + keytool -importkeystore -srckeystore server.p12 -destkeystore gateway.jks -srcstoretype pkcs12 + +While using this approach a couple of important things to be aware of: + +1. the alias MUST be "gateway-identity". You may need to change it using keytool after the import of the PKCS12 store. You can use keytool to do this - for example: + + keytool -changealias -alias "1" -destalias "gateway-identity" -keystore gateway.jks -storepass {knoxpw} + +2. the name of the expected identity keystore for the gateway MUST be `gateway.jks` +3. the passwords for the keystore and the imported key may both be set to the master secret for the gateway install. You can change the key passphrase after import using keytool as well. You may need to do this in order to provision the password in the credential store as described later in this section. For example: + + keytool -keypasswd -alias gateway-identity -keystore gateway.jks + +NOTE: The password for the keystore as well as that of the imported key may be the master secret for the gateway instance or you may set the gateway-identity-passphrase alias using the Knox CLI to the actual key passphrase. See the Knox CLI section for details. + +The following will allow you to provision the passphrase for the private key that you set during keystore creation above - it will prompt you for the actual passphrase. + + bin/knoxcli.sh create-alias gateway-identity-passphrase + +##### Generating a self-signed cert for use in testing or development environments ##### + + keytool -genkey -keyalg RSA -alias gateway-identity -keystore gateway.jks \ + -storepass {master-secret} -validity 360 -keysize 2048 + +Keytool will prompt you for a number of elements used will comprise the distinguished name (DN) within your certificate. + +*NOTE:* When it prompts you for your First and Last name be sure to type in the hostname of the machine that your gateway instance will be running on. This is used by clients during hostname verification to ensure that the presented certificate matches the hostname that was used in the URL for the connection - so they need to match. + +*NOTE:* When it prompts for the key password just press enter to ensure that it is the same as the keystore password. Which, as was described earlier, must match the master secret for the gateway instance. Alternatively, you can set it to another passphrase - take note of it and set the gateway-identity-passphrase alias to that passphrase using the Knox CLI. + +See the Knox CLI section for descriptions of the command line utilities related to the management of the keystores. + +##### Using a CA Signed Key Pair ##### +For certain deployments a certificate key pair that is signed by a trusted certificate authority is required. There are a number of different ways in which these certificates are acquired and can be converted and imported into the Apache Knox keystore. + +The following steps have been used to do this and are provided here for guidance in your installation. +You may have to adjust according to your environment. + +General steps: + +1. Stop Knox gateway and back up all files in `{GATEWAY_HOME}/data/security/keystores` + + gateway.sh stop + +2. Create a new master key for Knox and persist it. The master key will be referred to in following steps as `$master-key` + + knoxcli.sh create-master -force + +3. Create identity keystore gateway.jks. cert in alias gateway-identity + + cd {GATEWAY_HOME}/data/security/keystore + keytool -genkeypair -alias gateway-identity -keyalg RSA -keysize 1024 -dname "CN=$fqdn_knox,OU=hdp,O=sdge" -keypass $keypass -keystore gateway.jks -storepass $master-key -validity 300 + + NOTE: `$fqdn_knox` is the hostname of the Knox host. Some may choose `$keypass` to be the same as `$master-key`. + +4. Create credential store to store the `$keypass` in step 3. This creates `__gateway-credentials.jceks` file + + knoxcli.sh create-alias gateway-identity-passphrase --value $keypass + +5. Generate a certificate signing request from the gateway.jks + + keytool -keystore gateway.jks -storepass $master-key -alias gateway-identity -certreq -file knox.csr + +6. Send the `knox.csr` file to the CA authority and get back the signed certificate (`knox.signed`). You also need the CA certificate, which normally can be requested through an openssl command or web browser or from the CA. + +7. Import both the CA authority certificate (referred as `corporateCA.cer`) and the signed Knox certificate back into `gateway.jks` + + keytool -keystore gateway.jks -storepass $master-key -alias $hwhq -import -file corporateCA.cer + keytool -keystore gateway.jks -storepass $master-key -alias gateway-identity -import -file knox.signed + + NOTE: Use any alias appropriate for the corporate CA. + +8. Restart Knox gateway. Check `gateway.log` to check whether the gateway started properly and clusters are deployed. You can check the timestamp on cluster deployment files + + ls -alrt {GATEWAY_HOME}/data/deployment + +9. Verify that clients can use the CA authority cert to access Knox (which is the goal of using public signed cert) using curl or a web browsers which has the CA certificate installed + + curl --cacert supwin12ad.cer -u hdptester:hadoop -X GET 'https://$fqdn_knox:8443/gateway/$topologyname/webhdfs/v1/tmp?op=LISTSTATUS' + +##### Credential Store ##### +Whenever you provide your own keystore with either a self-signed cert or an issued certificate signed by a trusted authority, you will need to set an alias for the `gateway-identity-passphrase` or create an empty credential store. This is necessary for the current release in order for the system to determine the correct password for the keystore and the key. + +The credential stores in Knox use the JCEKS keystore type as it allows for the storage of general secrets in addition to certificates. + +Keytool may be used to create credential stores but the Knox CLI section details how to create aliases. These aliases are managed within credential stores which are created by the CLI as needed. The simplest approach is to create the `gateway-identity-passphrase` alias with the Knox CLI. This will create the credential store if it doesn't already exist and add the key passphrase. + +See the Knox CLI section for descriptions of the command line utilities related to the management of the credential stores. + +##### Provisioning of Keystores ##### +Once you have created these keystores you must move them into place for the gateway to discover them and use them to represent its identity for SSL connections. This is done by copying the keystores to the `{GATEWAY_HOME}/data/security/keystores` directory for your gateway install. + +#### Summary of Secrets to be Managed #### + +1. Master secret - the same for all gateway instances in a cluster of gateways +2. All security related artifacts are protected with the master secret +3. Secrets used by the gateway itself are stored within the gateway credential store and are the same across all gateway instances in the cluster of gateways +4. Secrets used by providers within cluster topologies are stored in topology specific credential stores and are the same for the same topology across the cluster of gateway instances. + However, they are specific to the topology - so secrets for one Hadoop cluster are different from those of another. + This allows for fail-over from one gateway instance to another even when encryption is being used while not allowing the compromise of one encryption key to expose the data for all clusters. + +NOTE: the SSL certificate will need special consideration depending on the type of certificate. Wildcard certs may be able to be shared across all gateway instances in a cluster. +When certs are dedicated to specific machines the gateway identity store will not be able to be blindly replicated as host name verification problems will ensue. +Obviously, trust-stores will need to be taken into account as well.
Added: knox/trunk/books/1.3.0/config_advanced_ldap.md URL: http://svn.apache.org/viewvc/knox/trunk/books/1.3.0/config_advanced_ldap.md?rev=1850181&view=auto ============================================================================== --- knox/trunk/books/1.3.0/config_advanced_ldap.md (added) +++ knox/trunk/books/1.3.0/config_advanced_ldap.md Wed Jan 2 17:31:29 2019 @@ -0,0 +1,281 @@ +### Advanced LDAP Authentication + +The default configuration computes the bind DN for incoming user based on userDnTemplate. +This does not work in enterprises where users could belong to multiple branches of LDAP tree. +You could instead enable advanced configuration that would compute bind DN of incoming user with an LDAP search. + +#### Problem with userDnTemplate based Authentication + +UserDnTemplate based authentication uses the configuration parameter `ldapRealm.userDnTemplate`. +Typical values of userDNTemplate would look like `uid={0},ou=people,dc=hadoop,dc=apache,dc=org`. + +To compute bind DN of the client, we swap the place holder `{0}` with the login id provided by the client. +For example, if the login id provided by the client is "guest', +the computed bind DN would be `uid=guest,ou=people,dc=hadoop,dc=apache,dc=org`. + +This keeps configuration simple. + +However, this does not work if users belong to different branches of LDAP DIT. +For example, if there are some users under `ou=people,dc=hadoop,dc=apache,dc=org` +and some users under `ou=contractors,dc=hadoop,dc=apache,dc=org`, +we cannot come up with userDnTemplate that would work for all the users. + +#### Using advanced LDAP Authentication + +With advanced LDAP authentication, we find the bind DN of the user by searching the LDAP directory +instead of interpolating the bind DN from userDNTemplate. + + +#### Example search filter to find the client bind DN + +Assuming + +* ldapRealm.userSearchAttributeName=uid +* ldapRealm.userObjectClass=person +* client specified login id = "guest" + +The LDAP Filter for doing a search to find the bind DN would be + + (&(uid=guest)(objectclass=person)) + +This could find the bind DN to be + + uid=guest,ou=people,dc=hadoop,dc=apache,dc=org + +Please note that the `userSearchAttributeName` need not be part of bindDN. + +For example, you could use + +* ldapRealm.userSearchAttributeName=email +* ldapRealm.userObjectClass=person +* client specified login id = "bill.clin...@gmail.com" + +The LDAP Filter for doing a search to find the bind DN would be + + (&(email=bill.clin...@gmail.com)(objectclass=person)) + +This could find the bind DN to be + + uid=billc,ou=contractors,dc=hadoop,dc=apache,dc=org + +#### Advanced LDAP configuration parameters +The table below provides a brief description and sample of the available advanced bind and search configuration parameters. + +| Parameter | Description | Default | Sample | +|-----------------------------|----------------------------------------------------------------|---------|--------------------------------------------------------------------| +| principalRegex | Parses the principal for insertion into templates via regex. | (.*) | (.\*?)\\\\(.\*) _(e.g. match US\tom: {0}=US\tom, {1}=US, {2}=tom)_ | +| userDnTemplate | Direct user bind DN template. | {0} | cn={2},dc={1},dc=qa,dc=company,dc=com | +| userSearchBase | Search based template. Used with config below. | none | dc={1},dc=qa,dc=company,dc=com | +| userSearchAttributeName | Attribute name for simplified search filter. | none | sAMAccountName | +| userSearchAttributeTemplate | Attribute template for simplified search filter. | {0} | {2} | +| userSearchFilter | Advanced search filter template. Note \& is \& in XML. | none | (\&(objectclass=person)(sAMAccountName={2})) | +| userSearchScope | Search scope: subtree, onelevel, object. | subtree | onelevel | + +#### Advanced LDAP configuration combinations +There are also only certain valid combinations of advanced LDAP configuration parameters. + +* User DN Template + * userDnTemplate (Required) + * principalRegex (Optional) +* User Search by Attribute + * userSearchBase (Required) + * userAttributeName (Required) + * userAttributeTemplate (Optional) + * userSearchScope (Optional) + * principalRegex (Optional) +* User Search by Filter + * userSearchBase (Required) + * userSearchFilter (Required) + * userSearchScope (Optional) + * principalRegex (Optional) + +#### Advanced LDAP configuration precedence +The presence of multiple configuration combinations should be avoided. +The rules below clarify which combinations take precedence when present. + +1. userSearchBase takes precedence over userDnTemplate +2. userSearchFilter takes precedence over userSearchAttributeName + +#### Example provider configuration to use advanced LDAP authentication + +The example configuration appears verbose due to the presence of liberal comments +and illustration of optional parameters and default values. +The configuration that you would use could be much shorter if you rely on default values. + + <provider> + + <role>authentication</role> + <name>ShiroProvider</name> + <enabled>true</enabled> + + <param> + <name>main.ldapRealm</name> + <value>org.apache.knox.gateway.shirorealm.KnoxLdapRealm</value> + </param> + + <param> + <name>main.ldapContextFactory</name> + <value>org.apache.knox.gateway.shirorealm.KnoxLdapContextFactory</value> + </param> + + <param> + <name>main.ldapRealm.contextFactory</name> + <value>$ldapContextFactory</value> + </param> + + <!-- update the value based on your ldap directory protocol, host and port --> + <param> + <name>main.ldapRealm.contextFactory.url</name> + <value>ldap://hdp.example.com:389</value> + </param> + + <!-- optional, default value: simple + Update the value based on mechanisms supported by your ldap directory --> + <param> + <name>main.ldapRealm.contextFactory.authenticationMechanism</name> + <value>simple</value> + </param> + + <!-- optional, default value: {0} + update the value based on your ldap DIT(directory information tree). + ignored if value is defined for main.ldapRealm.userSearchAttributeName --> + <param> + <name>main.ldapRealm.userDnTemplate</name> + <value>uid={0},ou=people,dc=hadoop,dc=apache,dc=org</value> + </param> + + <!-- optional, default value: null + If you specify a value for this attribute, useDnTemplate + specified above would be ignored and user bind DN would be computed using + ldap search + update the value based on your ldap DIT(directory information layout) + value of search attribute should identity the user uniquely --> + <param> + <name>main.ldapRealm.userSearchAttributeName</name> + <value>uid</value> + </param> + + <!-- optional, default value: false + If the value is true, groups in which user is a member are looked up + from LDAP and made available for service level authorization checks --> + <param> + <name>main.ldapRealm.authorizationEnabled</name> + <value>true</value> + </param> + + <!-- bind DN used to search for groups and user bind DN. + Required if a value is defined for main.ldapRealm.userSearchAttributeName + or if the value of main.ldapRealm.authorizationEnabled is true --> + <param> + <name>main.ldapRealm.contextFactory.systemUsername</name> + <value>uid=guest,ou=people,dc=hadoop,dc=apache,dc=org</value> + </param> + + <!-- password for systemUserName. + Required if a value is defined for main.ldapRealm.userSearchAttributeName + or if the value of main.ldapRealm.authorizationEnabled is true --> + <param> + <name>main.ldapRealm.contextFactory.systemPassword</name> + <value>${ALIAS=ldcSystemPassword}</value> + </param> + + <!-- optional, default value: simple + Update the value based on mechanisms supported by your ldap directory --> + <param> + <name>main.ldapRealm.contextFactory.systemAuthenticationMechanism</name> + <value>simple</value> + </param> + + <!-- optional, default value: person + Objectclass to identify user entries in ldap, used to build search + filter to search for user bind DN --> + <param> + <name>main.ldapRealm.userObjectClass</name> + <value>person</value> + </param> + + <!-- search base used to search for user bind DN and groups --> + <param> + <name>main.ldapRealm.searchBase</name> + <value>dc=hadoop,dc=apache,dc=org</value> + </param> + + <!-- search base used to search for user bind DN. + Defaults to the value of main.ldapRealm.searchBase. + If main.ldapRealm.userSearchAttributeName is defined, + value for main.ldapRealm.searchBase or main.ldapRealm.userSearchBase + should be defined --> + <param> + <name>main.ldapRealm.userSearchBase</name> + <value>dc=hadoop,dc=apache,dc=org</value> + </param> + + <!-- search base used to search for groups. + Defaults to the value of main.ldapRealm.searchBase. + If value of main.ldapRealm.authorizationEnabled is true, + value for main.ldapRealm.searchBase or main.ldapRealm.groupSearchBase should be defined --> + <param> + <name>main.ldapRealm.groupSearchBase</name> + <value>dc=hadoop,dc=apache,dc=org</value> + </param> + + <!-- optional, default value: groupOfNames + Objectclass to identify group entries in ldap, used to build search + filter to search for group entries --> + <param> + <name>main.ldapRealm.groupObjectClass</name> + <value>groupOfNames</value> + </param> + + <!-- optional, default value: member + If value is memberUrl, we treat found groups as dynamic groups --> + <param> + <name>main.ldapRealm.memberAttribute</name> + <value>member</value> + </param> + + <!-- optional, default value: uid={0} + Ignored if value is defined for main.ldapRealm.userSearchAttributeName --> + <param> + <name>main.ldapRealm.memberAttributeValueTemplate</name> + <value>uid={0},ou=people,dc=hadoop,dc=apache,dc=org</value> + </param> + + <!-- optional, default value: cn --> + <param> + <name>main.ldapRealm.groupIdAttribute</name> + <value>cn</value> + </param> + + <param> + <name>urls./**</name> + <value>authcBasic</value> + </param> + + <!-- optional, default value: 30min --> + <param> + <name>sessionTimeout</name> + <value>30</value> + </param> + + </provider> + +#### Special note on parameter main.ldapRealm.contextFactory.systemPassword + +The value for this could have one of the following 2 formats + +* plaintextpassword +* ${ALIAS=ldcSystemPassword} + +The first format specifies the password in plain text in the provider configuration. +Use of this format should be limited for testing and troubleshooting. + +We strongly recommend using the second format `${ALIAS=ldcSystemPassword}` in production. +This format uses an alias for the password stored in credential store. +In the example `${ALIAS=ldcSystemPassword}`, +ldcSystemPassword is the alias for the password stored in credential store. + +Assuming the plain text password is "hadoop", and your topology file name is "hdp.xml", +you would use following command to create the right password alias in credential store. + + {GATEWAY_HOME}/bin/knoxcli.sh create-alias ldcSystemPassword --cluster hdp --value hadoop Added: knox/trunk/books/1.3.0/config_audit.md URL: http://svn.apache.org/viewvc/knox/trunk/books/1.3.0/config_audit.md?rev=1850181&view=auto ============================================================================== --- knox/trunk/books/1.3.0/config_audit.md (added) +++ knox/trunk/books/1.3.0/config_audit.md Wed Jan 2 17:31:29 2019 @@ -0,0 +1,72 @@ +<!--- + 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. +---> + +### Audit ### + +The Audit facility within the Knox Gateway introduces functionality for tracking actions that are executed by Knox per user's request or that are produced by Knox internal events like topology deploy, etc. +The Knox Audit module is based on [Apache log4j](http://logging.apache.org/log4j/1.2/). + +#### Configuration needed #### + +Out of the box, the Knox Gateway includes preconfigured auditing capabilities. To change its configuration please read the following sections. + +#### Where audit logs go #### + +The Audit module is preconfigured to write audit records to the log file `{GATEWAY_HOME}/log/gateway-audit.log`. + +This behavior can be changed in the `{GATEWAY_HOME}/conf/gateway-log4j.properties` file. `app.audit.file` can be used to change the location. The `log4j.appender.auditfile.*` properties can be used for further customization. For detailed information read the [Apache log4j](http://logging.apache.org/log4j/1.2/) documentation. + +#### Audit format #### + +Out of the box, the audit record format is defined by `org.apache.knox.gateway.audit.log4j.layout.AuditLayout`. +Its structure is as follows: + + EVENT_PUBLISHING_TIME ROOT_REQUEST_ID|PARENT_REQUEST_ID|REQUEST_ID|LOGGER_NAME|TARGET_SERVICE_NAME|USER_NAME|PROXY_USER_NAME|SYSTEM_USER_NAME|ACTION|RESOURCE_TYPE|RESOURCE_NAME|OUTCOME|LOGGING_MESSAGE + +The audit record format can be changed by setting `log4j.appender.auditfile.layout` property in `{GATEWAY_HOME}/conf/gateway-log4j.properties` to another class that extends `org.apache.log4j.Layout` or its subclasses. + +For detailed information read [Apache log4j](http://logging.apache.org/log4j/1.2/). + +##### How to interpret audit log ##### + +Component | Description +---------|----------- +EVENT_PUBLISHING_TIME | Time when audit record was published. +ROOT_REQUEST_ID | The root request ID if this is a sub-request. Currently it is empty. +PARENT_REQUEST_ID | The parent request ID if this is a sub-request. Currently it is empty. +REQUEST_ID | A unique value representing the current, active request. If the current request id value is different from the current parent request id value then the current request id value is moved to the parent request id before it is replaced by the provided request id. If the root request id is not set it will be set with the first non-null value of either the parent request id or the passed request id. +LOGGER_NAME | The name of the logger +TARGET_SERVICE_NAME | Name of Hadoop service. Can be empty if audit record is not linked to any Hadoop service, for example, audit record for topology deployment. +USER_NAME | Name of user that initiated session with Knox +PROXY_USER_NAME | Mapped user name. For detailed information read #[Identity Assertion]. +SYSTEM_USER_NAME | Currently is empty. +ACTION | Type of action that was executed. Following actions are defined: authentication, authorization, redeploy, deploy, undeploy, identity-mapping, dispatch, access. +RESOURCE_TYPE | Type of resource for which action was executed. Following resource types are defined: uri, topology, principal. +RESOURCE_NAME | Name of resource. For resource of type topology it is name of topology. For resource of type uri it is inbound or dispatch request path. For resource of type principal it is a name of mapped user. +OUTCOME | Action result type. Following outcomes are defined: success, failure, unavailable. +LOGGING_MESSAGE | Logging message. Contains additional tracking information. + +#### Audit log rotation #### + +Audit logging is preconfigured with `org.apache.log4j.DailyRollingFileAppender`. +[Apache log4j](http://logging.apache.org/log4j/1.2/) contains information about other Appenders. + +#### How to change the audit level or disable it #### + +All audit messages are logged at `INFO` level and this behavior can't be changed. + +Disabling auditing can be done by decreasing the log level for the Audit appender or setting it to `OFF`. Added: knox/trunk/books/1.3.0/config_authn.md URL: http://svn.apache.org/viewvc/knox/trunk/books/1.3.0/config_authn.md?rev=1850181&view=auto ============================================================================== --- knox/trunk/books/1.3.0/config_authn.md (added) +++ knox/trunk/books/1.3.0/config_authn.md Wed Jan 2 17:31:29 2019 @@ -0,0 +1,159 @@ +<!--- + 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. +---> + +### Authentication ### + +There are two types of providers supported in Knox for establishing a user's identity: + +1. Authentication Providers +2. Federation Providers + +Authentication providers directly accept a user's credentials and validates them against some particular user store. Federation providers, on the other hand, validate a token that has been issued for the user by a trusted Identity Provider (IdP). + +The current release of Knox ships with an authentication provider based on the Apache Shiro project and is initially configured for BASIC authentication against an LDAP store. This has been specifically tested against Apache Directory Server and Active Directory. + +This section will cover the general approach to leveraging Shiro within the bundled provider including: + +1. General mapping of provider config to `shiro.ini` config +2. Specific configuration for the bundled BASIC/LDAP configuration +3. Some tips into what may need to be customized for your environment +4. How to setup the use of LDAP over SSL or LDAPS + +#### General Configuration for Shiro Provider #### + +As is described in the configuration section of this document, providers have a name-value based configuration - as is the common pattern in the rest of Hadoop. + +The following example shows the format of the configuration for a given provider: + + <provider> + <role>authentication</role> + <name>ShiroProvider</name> + <enabled>true</enabled> + <param> + <name>{name}</name> + <value>{value}</value> + </param> + </provider> + +Conversely, the Shiro provider currently expects a `shiro.ini` file in the `WEB-INF` directory of the cluster specific web application. + +The following example illustrates a configuration of the bundled BASIC/LDAP authentication config in a `shiro.ini` file: + + [urls] + /**=authcBasic + [main] + ldapRealm=org.apache.shiro.realm.ldap.JndiLdapRealm + ldapRealm.contextFactory.authenticationMechanism=simple + ldapRealm.contextFactory.url=ldap://localhost:33389 + ldapRealm.userDnTemplate=uid={0},ou=people,dc=hadoop,dc=apache,dc=org + +In order to fit into the context of an INI file format, at deployment time we interrogate the parameters provided in the provider configuration and parse the INI section out of the parameter names. The following provider config illustrates this approach. Notice that the section names in the above shiro.ini match the beginning of the parameter names that are in the following config: + + <gateway> + <provider> + <role>authentication</role> + <name>ShiroProvider</name> + <enabled>true</enabled> + <param> + <name>main.ldapRealm</name> + <value>org.apache.shiro.realm.ldap.JndiLdapRealm</value> + </param> + <param> + <name>main.ldapRealm.userDnTemplate</name> + <value>uid={0},ou=people,dc=hadoop,dc=apache,dc=org</value> + </param> + <param> + <name>main.ldapRealm.contextFactory.url</name> + <value>ldap://localhost:33389</value> + </param> + <param> + <name>main.ldapRealm.contextFactory.authenticationMechanism</name> + <value>simple</value> + </param> + <param> + <name>urls./**</name> + <value>authcBasic</value> + </param> + </provider> + +This happens to be the way that we are currently configuring Shiro for BASIC/LDAP authentication. This same config approach may be used to achieve other authentication mechanisms or variations on this one. We however have not tested additional uses for it for this release. + +#### LDAP Configuration #### + +This section discusses the LDAP configuration used above for the Shiro Provider. Some of these configuration elements will need to be customized to reflect your deployment environment. + +**main.ldapRealm** - this element indicates the fully qualified class name of the Shiro realm to be used in authenticating the user. The class name provided by default in the sample is the `org.apache.shiro.realm.ldap.JndiLdapRealm` this implementation provides us with the ability to authenticate but by default has authorization disabled. In order to provide authorization - which is seen by Shiro as dependent on an LDAP schema that is specific to each organization - an extension of JndiLdapRealm is generally used to override and implement the doGetAuthorizationInfo method. In this particular release we are providing a simple authorization provider that can be used along with the Shiro authentication provider. + +**main.ldapRealm.userDnTemplate** - in order to bind a simple username to an LDAP server that generally requires a full distinguished name (DN), we must provide the template into which the simple username will be inserted. This template allows for the creation of a DN by injecting the simple username into the common name (CN) portion of the DN. **This element will need to be customized to reflect your deployment environment.** The template provided in the sample is only an example and is valid only within the LDAP schema distributed with Knox and is represented by the `users.ldif` file in the `{GATEWAY_HOME}/conf` directory. + +**main.ldapRealm.contextFactory.url** - this element is the URL that represents the host and port of the LDAP server. It also includes the scheme of the protocol to use. This may be either `ldap` or `ldaps` depending on whether you are communicating with the LDAP over SSL (highly recommended). **This element will need to be customized to reflect your deployment environment.**. + +**main.ldapRealm.contextFactory.authenticationMechanism** - this element indicates the type of authentication that should be performed against the LDAP server. The current default value is `simple` which indicates a simple bind operation. This element should not need to be modified and no mechanism other than a simple bind has been tested for this particular release. + +**urls./**** - this element represents a single URL_Ant_Path_Expression and the value the Shiro filter chain to apply to it. This particular sample indicates that all paths into the application have the same Shiro filter chain applied. The paths are relative to the application context path. The use of the value `authcBasic` here indicates that BASIC authentication is expected for every path into the application. Adding an additional Shiro filter to that chain for validating that the request isSecure() and over SSL can be achieved by changing the value to `ssl, authcBasic`. It is not likely that you need to change this element for your environment. + +#### Active Directory - Special Note #### + +You would use LDAP configuration as documented above to authenticate against Active Directory as well. + +Some Active Directory specific things to keep in mind: + +Typical AD `main.ldapRealm.userDnTemplate` value looks slightly different, such as + + cn={0},cn=users,DC=lab,DC=sample,dc=com + +Please compare this with a typical Apache DS `main.ldapRealm.userDnTemplate` value and make note of the difference: + + `uid={0},ou=people,dc=hadoop,dc=apache,dc=org` + +If your AD is configured to authenticate based on just the cn and password and does not require user DN, you do not have to specify value for `main.ldapRealm.userDnTemplate`. + + +#### LDAP over SSL (LDAPS) Configuration #### +In order to communicate with your LDAP server over SSL (again, highly recommended), you will need to modify the topology file in a couple ways and possibly provision some keying material. + +1. **main.ldapRealm.contextFactory.url** must be changed to have the `ldaps` protocol scheme and the port must be the SSL listener port on your LDAP server. +2. Identity certificate (keypair) provisioned to LDAP server - your LDAP server specific documentation should indicate what is required for providing a cert or keypair to represent the LDAP server identity to connecting clients. +3. Trusting the LDAP Server's public key - if the LDAP Server's identity certificate is issued by a well known and trusted certificate authority and is already represented in the JRE's cacerts truststore then you don't need to do anything for trusting the LDAP server's cert. If, however, the cert is self-signed or issued by an untrusted authority you will need to either add it to the cacerts keystore or to another truststore that you may direct Knox to utilize through a system property (`javax.net.ssl.trustStore` and `javax.net.ssl.trustStorePassword`). + +#### Session Configuration #### + +Knox maps each cluster topology to a web application and leverages standard JavaEE session management. + +To configure session idle timeout for the topology, please specify value of parameter sessionTimeout for ShiroProvider in your topology file. If you do not specify the value for this parameter, it defaults to 30 minutes. + +The definition would look like the following in the topology file: + + ... + <provider> + <role>authentication</role> + <name>ShiroProvider</name> + <enabled>true</enabled> + <param> + <!-- + Session timeout in minutes. This is really idle timeout. + Defaults to 30 minutes, if the property value is not defined. + Current client authentication will expire if client idles + continuously for more than this value + --> + <name>sessionTimeout</name> + <value>30</value> + </param> + <provider> + ... + +At present, ShiroProvider in Knox leverages JavaEE session to maintain authentication state for a user across requests using JSESSIONID cookie. So, a client that authenticated with Knox could pass the JSESSIONID cookie with repeated requests as long as the session has not timed out instead of submitting userid/password with every request. Presenting a valid session cookie in place of userid/password would also perform better as additional credential store lookups are avoided.
