问题导读
1.如何设置storm内存?2.如果没有配置文件的情况下,该如何配置一些参数?3.通过哪个参数可以配置内存?Storm中真正干活的是各个worker,而worker由supervisor负责启动。在topology启动过程中我们会看到如下的启动日志: 这就是启动一个worker进程,也就是一个JVM进程。默认情况下,Storm启动worker进程时,JVM的最大内存是768M。但我在使用过程中,由于会在Bolt中加载大量数据,768M内存无法满足需求,会导致内存溢出程序崩溃。经过研究发现,可以通过在Strom的配置文件storm.yaml中设置worker的启动参数:1 worker.childopts: "-Xmx2048m"
该参数会在启动时传递给JVM,然后就可以在worker中使用2048m内存了。目前好像Storm还没有配置文件的详细说明,比如可以配置哪些参数,怎么配置?大家可以先参考Storm源代码中的Config.java.1 package backtype.storm; 2 3 import backtype.storm.ConfigValidation; 4 import backtype.storm.serialization.IKryoDecorator; 5 import backtype.storm.serialization.IKryoFactory; 6 import com.esotericsoftware.kryo.Serializer; 7 import java.util.ArrayList; 8 import java.util.HashMap; 9 import java.util.List; 10 import java.util.Map; 11 12 /** 13 * Topology configs are specified as a plain old map. This class provides a 14 * convenient way to create a topology config map by providing setter methods for 15 * all the configs that can be set. It also makes it easier to do things like add 16 * serializations. 17 * 18 *This class also provides constants for all the configurations possible on 19 * a Storm cluster and Storm topology. Each constant is paired with a schema 20 * that defines the validity criterion of the corresponding field. Default 21 * values for these configs can be found in defaults.yaml.
22 * 23 *Note that you may put other configurations in any of the configs. Storm 24 * will ignore anything it doesn't recognize, but your topologies are free to make 25 * use of them by reading them in the prepare method of Bolts or the open method of 26 * Spouts.
27 */ 28 public class Config extends HashMap{ 29 /** 30 * The transporter for communication among Storm tasks 31 */ 32 public static final String STORM_MESSAGING_TRANSPORT = "storm.messaging.transport"; 33 public static final Object STORM_MESSAGING_TRANSPORT_SCHEMA = String.class; 34 35 /** 36 * Netty based messaging: The buffer size for send/recv buffer 37 */ 38 public static final String STORM_MESSAGING_NETTY_BUFFER_SIZE = "storm.messaging.netty.buffer_size"; 39 public static final Object STORM_MESSAGING_NETTY_BUFFER_SIZE_SCHEMA = Number.class; 40 41 /** 42 * Netty based messaging: The max # of retries that a peer will perform when a remote is not accessible 43 */ 44 public static final String STORM_MESSAGING_NETTY_MAX_RETRIES = "storm.messaging.netty.max_retries"; 45 public static final Object STORM_MESSAGING_NETTY_MAX_RETRIES_SCHEMA = Number.class; 46 47 /** 48 * Netty based messaging: The min # of milliseconds that a peer will wait. 49 */ 50 public static final String STORM_MESSAGING_NETTY_MIN_SLEEP_MS = "storm.messaging.netty.min_wait_ms"; 51 public static final Object STORM_MESSAGING_NETTY_MIN_SLEEP_MS_SCHEMA = Number.class; 52 53 /** 54 * Netty based messaging: The max # of milliseconds that a peer will wait. 55 */ 56 public static final String STORM_MESSAGING_NETTY_MAX_SLEEP_MS = "storm.messaging.netty.max_wait_ms"; 57 public static final Object STORM_MESSAGING_NETTY_MAX_SLEEP_MS_SCHEMA = Number.class; 58 59 /** 60 * Netty based messaging: The # of worker threads for the server. 61 */ 62 public static final String STORM_MESSAGING_NETTY_SERVER_WORKER_THREADS = "storm.messaging.netty.server_worker_threads"; 63 public static final Object STORM_MESSAGING_NETTY_SERVER_WORKER_THREADS_SCHEMA = Number.class; 64 65 /** 66 * Netty based messaging: The # of worker threads for the client. 67 */ 68 public static final String STORM_MESSAGING_NETTY_CLIENT_WORKER_THREADS = "storm.messaging.netty.client_worker_threads"; 69 public static final Object STORM_MESSAGING_NETTY_CLIENT_WORKER_THREADS_SCHEMA = Number.class; 70 71 /** 72 * A list of hosts of ZooKeeper servers used to manage the cluster. 73 */ 74 public static final String STORM_ZOOKEEPER_SERVERS = "storm.zookeeper.servers"; 75 public static final Object STORM_ZOOKEEPER_SERVERS_SCHEMA = ConfigValidation.StringsValidator; 76 77 /** 78 * The port Storm will use to connect to each of the ZooKeeper servers. 79 */ 80 public static final String STORM_ZOOKEEPER_PORT = "storm.zookeeper.port"; 81 public static final Object STORM_ZOOKEEPER_PORT_SCHEMA = Number.class; 82 83 /** 84 * A directory on the local filesystem used by Storm for any local 85 * filesystem usage it needs. The directory must exist and the Storm daemons must 86 * have permission to read/write from this location. 87 */ 88 public static final String STORM_LOCAL_DIR = "storm.local.dir"; 89 public static final Object STORM_LOCAL_DIR_SCHEMA = String.class; 90 91 /** 92 * A global task scheduler used to assign topologies's tasks to supervisors' wokers. 93 * 94 * If this is not set, a default system scheduler will be used. 95 */ 96 public static final String STORM_SCHEDULER = "storm.scheduler"; 97 public static final Object STORM_SCHEDULER_SCHEMA = String.class; 98 99 /**100 * The mode this Storm cluster is running in. Either "distributed" or "local".101 */102 public static final String STORM_CLUSTER_MODE = "storm.cluster.mode";103 public static final Object STORM_CLUSTER_MODE_SCHEMA = String.class;104 105 /**106 * The hostname the supervisors/workers should report to nimbus. If unset, Storm will 107 * get the hostname to report by calling InetAddress.getLocalHost().getCanonicalHostName()
.108 * 109 * You should set this config when you dont have a DNS which supervisors/workers110 * can utilize to find each other based on hostname got from calls to111 *InetAddress.getLocalHost().getCanonicalHostName()
.112 */113 public static final String STORM_LOCAL_HOSTNAME = "storm.local.hostname";114 public static final Object STORM_LOCAL_HOSTNAME_SCHEMA = String.class;115 116 /**117 * The transport plug-in for Thrift client/server communication118 */119 public static final String STORM_THRIFT_TRANSPORT_PLUGIN = "storm.thrift.transport";120 public static final Object STORM_THRIFT_TRANSPORT_PLUGIN_SCHEMA = String.class;121 122 /**123 * The serializer class for ListDelegate (tuple payload). 124 * The default serializer will be ListDelegateSerializer125 */126 public static final String TOPOLOGY_TUPLE_SERIALIZER = "topology.tuple.serializer";127 public static final Object TOPOLOGY_TUPLE_SERIALIZER_SCHEMA = String.class;128 129 /**130 * Whether or not to use ZeroMQ for messaging in local mode. If this is set 131 * to false, then Storm will use a pure-Java messaging system. The purpose 132 * of this flag is to make it easy to run Storm in local mode by eliminating 133 * the need for native dependencies, which can be difficult to install.134 *135 * Defaults to false.136 */137 public static final String STORM_LOCAL_MODE_ZMQ = "storm.local.mode.zmq";138 public static final Object STORM_LOCAL_MODE_ZMQ_SCHEMA = Boolean.class;139 140 /**141 * The root location at which Storm stores data in ZooKeeper.142 */143 public static final String STORM_ZOOKEEPER_ROOT = "storm.zookeeper.root";144 public static final Object STORM_ZOOKEEPER_ROOT_SCHEMA = String.class;145 146 /**147 * The session timeout for clients to ZooKeeper.148 */149 public static final String STORM_ZOOKEEPER_SESSION_TIMEOUT = "storm.zookeeper.session.timeout";150 public static final Object STORM_ZOOKEEPER_SESSION_TIMEOUT_SCHEMA = Number.class;151 152 /**153 * The connection timeout for clients to ZooKeeper.154 */155 public static final String STORM_ZOOKEEPER_CONNECTION_TIMEOUT = "storm.zookeeper.connection.timeout";156 public static final Object STORM_ZOOKEEPER_CONNECTION_TIMEOUT_SCHEMA = Number.class;157 158 /**159 * The number of times to retry a Zookeeper operation.160 */161 public static final String STORM_ZOOKEEPER_RETRY_TIMES="storm.zookeeper.retry.times";162 public static final Object STORM_ZOOKEEPER_RETRY_TIMES_SCHEMA = Number.class;163 164 /**165 * The interval between retries of a Zookeeper operation.166 */167 public static final String STORM_ZOOKEEPER_RETRY_INTERVAL="storm.zookeeper.retry.interval";168 public static final Object STORM_ZOOKEEPER_RETRY_INTERVAL_SCHEMA = Number.class;169 170 /**171 * The ceiling of the interval between retries of a Zookeeper operation.172 */173 public static final String STORM_ZOOKEEPER_RETRY_INTERVAL_CEILING="storm.zookeeper.retry.intervalceiling.millis";174 public static final Object STORM_ZOOKEEPER_RETRY_INTERVAL_CEILING_SCHEMA = Number.class;175 176 /**177 * The Zookeeper authentication scheme to use, e.g. "digest". Defaults to no authentication.178 */179 public static final String STORM_ZOOKEEPER_AUTH_SCHEME="storm.zookeeper.auth.scheme";180 public static final Object STORM_ZOOKEEPER_AUTH_SCHEME_SCHEMA = String.class;181 182 /**183 * A string representing the payload for Zookeeper authentication. It gets serialized using UTF-8 encoding during authentication.184 */185 public static final String STORM_ZOOKEEPER_AUTH_PAYLOAD="storm.zookeeper.auth.payload";186 public static final Object STORM_ZOOKEEPER_AUTH_PAYLOAD_SCHEMA = String.class;187 188 /**189 * The id assigned to a running topology. The id is the storm name with a unique nonce appended.190 */191 public static final String STORM_ID = "storm.id";192 public static final Object STORM_ID_SCHEMA = String.class;193 194 /**195 * The host that the master server is running on.196 */197 public static final String NIMBUS_HOST = "nimbus.host";198 public static final Object NIMBUS_HOST_SCHEMA = String.class;199 200 /**201 * Which port the Thrift interface of Nimbus should run on. Clients should202 * connect to this port to upload jars and submit topologies.203 */204 public static final String NIMBUS_THRIFT_PORT = "nimbus.thrift.port";205 public static final Object NIMBUS_THRIFT_PORT_SCHEMA = Number.class;206 207 208 /**209 * This parameter is used by the storm-deploy project to configure the210 * jvm options for the nimbus daemon.211 */212 public static final String NIMBUS_CHILDOPTS = "nimbus.childopts";213 public static final Object NIMBUS_CHILDOPTS_SCHEMA = String.class;214 215 216 /**217 * How long without heartbeating a task can go before nimbus will consider the218 * task dead and reassign it to another location.219 */220 public static final String NIMBUS_TASK_TIMEOUT_SECS = "nimbus.task.timeout.secs";221 public static final Object NIMBUS_TASK_TIMEOUT_SECS_SCHEMA = Number.class;222 223 224 /**225 * How often nimbus should wake up to check heartbeats and do reassignments. Note226 * that if a machine ever goes down Nimbus will immediately wake up and take action.227 * This parameter is for checking for failures when there's no explicit event like that228 * occuring.229 */230 public static final String NIMBUS_MONITOR_FREQ_SECS = "nimbus.monitor.freq.secs";231 public static final Object NIMBUS_MONITOR_FREQ_SECS_SCHEMA = Number.class;232 233 /**234 * How often nimbus should wake the cleanup thread to clean the inbox.235 * @see NIMBUS_INBOX_JAR_EXPIRATION_SECS236 */237 public static final String NIMBUS_CLEANUP_INBOX_FREQ_SECS = "nimbus.cleanup.inbox.freq.secs";238 public static final Object NIMBUS_CLEANUP_INBOX_FREQ_SECS_SCHEMA = Number.class;239 240 /**241 * The length of time a jar file lives in the inbox before being deleted by the cleanup thread.242 *243 * Probably keep this value greater than or equal to NIMBUS_CLEANUP_INBOX_JAR_EXPIRATION_SECS.244 * Note that the time it takes to delete an inbox jar file is going to be somewhat more than245 * NIMBUS_CLEANUP_INBOX_JAR_EXPIRATION_SECS (depending on how often NIMBUS_CLEANUP_FREQ_SECS246 * is set to).247 * @see NIMBUS_CLEANUP_FREQ_SECS248 */249 public static final String NIMBUS_INBOX_JAR_EXPIRATION_SECS = "nimbus.inbox.jar.expiration.secs";250 public static final Object NIMBUS_INBOX_JAR_EXPIRATION_SECS_SCHEMA = Number.class;251 252 /**253 * How long before a supervisor can go without heartbeating before nimbus considers it dead254 * and stops assigning new work to it.255 */256 public static final String NIMBUS_SUPERVISOR_TIMEOUT_SECS = "nimbus.supervisor.timeout.secs";257 public static final Object NIMBUS_SUPERVISOR_TIMEOUT_SECS_SCHEMA = Number.class;258 259 /**260 * A special timeout used when a task is initially launched. During launch, this is the timeout261 * used until the first heartbeat, overriding nimbus.task.timeout.secs.262 *263 *A separate timeout exists for launch because there can be quite a bit of overhead264 * to launching new JVM's and configuring them.
265 */266 public static final String NIMBUS_TASK_LAUNCH_SECS = "nimbus.task.launch.secs";267 public static final Object NIMBUS_TASK_LAUNCH_SECS_SCHEMA = Number.class;268 269 /**270 * Whether or not nimbus should reassign tasks if it detects that a task goes down. 271 * Defaults to true, and it's not recommended to change this value.272 */273 public static final String NIMBUS_REASSIGN = "nimbus.reassign";274 public static final Object NIMBUS_REASSIGN_SCHEMA = Boolean.class;275 276 /**277 * During upload/download with the master, how long an upload or download connection is idle278 * before nimbus considers it dead and drops the connection.279 */280 public static final String NIMBUS_FILE_COPY_EXPIRATION_SECS = "nimbus.file.copy.expiration.secs";281 public static final Object NIMBUS_FILE_COPY_EXPIRATION_SECS_SCHEMA = Number.class;282 283 /**284 * A custom class that implements ITopologyValidator that is run whenever a285 * topology is submitted. Can be used to provide business-specific logic for286 * whether topologies are allowed to run or not.287 */288 public static final String NIMBUS_TOPOLOGY_VALIDATOR = "nimbus.topology.validator";289 public static final Object NIMBUS_TOPOLOGY_VALIDATOR_SCHEMA = String.class;290 291 /**292 * Class name for authorization plugin for Nimbus293 */294 public static final String NIMBUS_AUTHORIZER = "nimbus.authorizer";295 public static final Object NIMBUS_AUTHORIZER_SCHEMA = String.class;296 297 /**298 * Storm UI binds to this port.299 */300 public static final String UI_PORT = "ui.port";301 public static final Object UI_PORT_SCHEMA = Number.class;302 303 /**304 * HTTP UI port for log viewer305 */306 public static final String LOGVIEWER_PORT = "logviewer.port";307 public static final Object LOGVIEWER_PORT_SCHEMA = Number.class;308 309 /**310 * Childopts for log viewer java process.311 */312 public static final String LOGVIEWER_CHILDOPTS = "logviewer.childopts";313 public static final Object LOGVIEWER_CHILDOPTS_SCHEMA = String.class;314 315 /**316 * Appender name used by log viewer to determine log directory.317 */318 public static final String LOGVIEWER_APPENDER_NAME = "logviewer.appender.name";319 public static final Object LOGVIEWER_APPENDER_NAME_SCHEMA = String.class;320 321 /**322 * Childopts for Storm UI Java process.323 */324 public static final String UI_CHILDOPTS = "ui.childopts";325 public static final Object UI_CHILDOPTS_SCHEMA = String.class;326 327 /**328 * List of DRPC servers so that the DRPCSpout knows who to talk to.329 */330 public static final String DRPC_SERVERS = "drpc.servers";331 public static final Object DRPC_SERVERS_SCHEMA = ConfigValidation.StringsValidator;332 333 /**334 * This port is used by Storm DRPC for receiving DPRC requests from clients.335 */336 public static final String DRPC_PORT = "drpc.port";337 public static final Object DRPC_PORT_SCHEMA = Number.class;338 339 /**340 * DRPC thrift server worker threads 341 */342 public static final String DRPC_WORKER_THREADS = "drpc.worker.threads";343 public static final Object DRPC_WORKER_THREADS_SCHEMA = Number.class;344 345 /**346 * DRPC thrift server queue size 347 */348 public static final String DRPC_QUEUE_SIZE = "drpc.queue.size";349 public static final Object DRPC_QUEUE_SIZE_SCHEMA = Number.class;350 351 /**352 * This port on Storm DRPC is used by DRPC topologies to receive function invocations and send results back. 353 */354 public static final String DRPC_INVOCATIONS_PORT = "drpc.invocations.port";355 public static final Object DRPC_INVOCATIONS_PORT_SCHEMA = Number.class;356 357 /**358 * The timeout on DRPC requests within the DRPC server. Defaults to 10 minutes. Note that requests can also359 * timeout based on the socket timeout on the DRPC client, and separately based on the topology message360 * timeout for the topology implementing the DRPC function.361 */362 public static final String DRPC_REQUEST_TIMEOUT_SECS = "drpc.request.timeout.secs";363 public static final Object DRPC_REQUEST_TIMEOUT_SECS_SCHEMA = Number.class;364 365 /**366 * Childopts for Storm DRPC Java process.367 */368 public static final String DRPC_CHILDOPTS = "drpc.childopts";369 public static final Object DRPC_CHILDOPTS_SCHEMA = String.class;370 371 /**372 * the metadata configed on the supervisor373 */374 public static final String SUPERVISOR_SCHEDULER_META = "supervisor.scheduler.meta";375 public static final Object SUPERVISOR_SCHEDULER_META_SCHEMA = Map.class;376 /**377 * A list of ports that can run workers on this supervisor. Each worker uses one port, and378 * the supervisor will only run one worker per port. Use this configuration to tune379 * how many workers run on each machine.380 */381 public static final String SUPERVISOR_SLOTS_PORTS = "supervisor.slots.ports";382 public static final Object SUPERVISOR_SLOTS_PORTS_SCHEMA = ConfigValidation.NumbersValidator;383 384 385 /**386 * This parameter is used by the storm-deploy project to configure the387 * jvm options for the supervisor daemon.388 */389 public static final String SUPERVISOR_CHILDOPTS = "supervisor.childopts";390 public static final Object SUPERVISOR_CHILDOPTS_SCHEMA = String.class;391 392 393 /**394 * How long a worker can go without heartbeating before the supervisor tries to395 * restart the worker process.396 */397 public static final String SUPERVISOR_WORKER_TIMEOUT_SECS = "supervisor.worker.timeout.secs";398 public static final Object SUPERVISOR_WORKER_TIMEOUT_SECS_SCHEMA = Number.class;399 400 401 /**402 * How long a worker can go without heartbeating during the initial launch before403 * the supervisor tries to restart the worker process. This value override404 * supervisor.worker.timeout.secs during launch because there is additional405 * overhead to starting and configuring the JVM on launch.406 */407 public static final String SUPERVISOR_WORKER_START_TIMEOUT_SECS = "supervisor.worker.start.timeout.secs";408 public static final Object SUPERVISOR_WORKER_START_TIMEOUT_SECS_SCHEMA = Number.class;409 410 411 /**412 * Whether or not the supervisor should launch workers assigned to it. Defaults413 * to true -- and you should probably never change this value. This configuration414 * is used in the Storm unit tests.415 */416 public static final String SUPERVISOR_ENABLE = "supervisor.enable";417 public static final Object SUPERVISOR_ENABLE_SCHEMA = Boolean.class;418 419 420 /**421 * how often the supervisor sends a heartbeat to the master.422 */423 public static final String SUPERVISOR_HEARTBEAT_FREQUENCY_SECS = "supervisor.heartbeat.frequency.secs";424 public static final Object SUPERVISOR_HEARTBEAT_FREQUENCY_SECS_SCHEMA = Number.class;425 426 427 /**428 * How often the supervisor checks the worker heartbeats to see if any of them429 * need to be restarted.430 */431 public static final String SUPERVISOR_MONITOR_FREQUENCY_SECS = "supervisor.monitor.frequency.secs";432 public static final Object SUPERVISOR_MONITOR_FREQUENCY_SECS_SCHEMA = Number.class;433 434 /**435 * The jvm opts provided to workers launched by this supervisor. All "%ID%" substrings are replaced436 * with an identifier for this worker.437 */438 public static final String WORKER_CHILDOPTS = "worker.childopts";439 public static final Object WORKER_CHILDOPTS_SCHEMA = String.class;440 441 442 /**443 * How often this worker should heartbeat to the supervisor.444 */445 public static final String WORKER_HEARTBEAT_FREQUENCY_SECS = "worker.heartbeat.frequency.secs";446 public static final Object WORKER_HEARTBEAT_FREQUENCY_SECS_SCHEMA = Number.class;447 448 /**449 * How often a task should heartbeat its status to the master.450 */451 public static final String TASK_HEARTBEAT_FREQUENCY_SECS = "task.heartbeat.frequency.secs";452 public static final Object TASK_HEARTBEAT_FREQUENCY_SECS_SCHEMA = Number.class;453 454 455 /**456 * How often a task should sync its connections with other tasks (if a task is457 * reassigned, the other tasks sending messages to it need to refresh their connections).458 * In general though, when a reassignment happens other tasks will be notified459 * almost immediately. This configuration is here just in case that notification doesn't460 * come through.461 */462 public static final String TASK_REFRESH_POLL_SECS = "task.refresh.poll.secs";463 public static final Object TASK_REFRESH_POLL_SECS_SCHEMA = Number.class;464 465 466 467 /**468 * True if Storm should timeout messages or not. Defaults to true. This is meant to be used469 * in unit tests to prevent tuples from being accidentally timed out during the test.470 */471 public static final String TOPOLOGY_ENABLE_MESSAGE_TIMEOUTS = "topology.enable.message.timeouts";472 public static final Object TOPOLOGY_ENABLE_MESSAGE_TIMEOUTS_SCHEMA = Boolean.class;473 474 /**475 * When set to true, Storm will log every message that's emitted.476 */477 public static final String TOPOLOGY_DEBUG = "topology.debug";478 public static final Object TOPOLOGY_DEBUG_SCHEMA = Boolean.class;479 480 481 /**482 * Whether or not the master should optimize topologies by running multiple483 * tasks in a single thread where appropriate.484 */485 public static final String TOPOLOGY_OPTIMIZE = "topology.optimize";486 public static final Object TOPOLOGY_OPTIMIZE_SCHEMA = Boolean.class;487 488 /**489 * How many processes should be spawned around the cluster to execute this490 * topology. Each process will execute some number of tasks as threads within491 * them. This parameter should be used in conjunction with the parallelism hints492 * on each component in the topology to tune the performance of a topology.493 */494 public static final String TOPOLOGY_WORKERS = "topology.workers";495 public static final Object TOPOLOGY_WORKERS_SCHEMA = Number.class;496 497 /**498 * How many instances to create for a spout/bolt. A task runs on a thread with zero or more499 * other tasks for the same spout/bolt. The number of tasks for a spout/bolt is always500 * the same throughout the lifetime of a topology, but the number of executors (threads) for 501 * a spout/bolt can change over time. This allows a topology to scale to more or less resources 502 * without redeploying the topology or violating the constraints of Storm (such as a fields grouping503 * guaranteeing that the same value goes to the same task).504 */505 public static final String TOPOLOGY_TASKS = "topology.tasks";506 public static final Object TOPOLOGY_TASKS_SCHEMA = Number.class;507 508 /**509 * How many executors to spawn for ackers.510 *511 *If this is set to 0, then Storm will immediately ack tuples as soon512 * as they come off the spout, effectively disabling reliability.
513 */514 public static final String TOPOLOGY_ACKER_EXECUTORS = "topology.acker.executors";515 public static final Object TOPOLOGY_ACKER_EXECUTORS_SCHEMA = Number.class;516 517 518 /**519 * The maximum amount of time given to the topology to fully process a message520 * emitted by a spout. If the message is not acked within this time frame, Storm521 * will fail the message on the spout. Some spouts implementations will then replay522 * the message at a later time.523 */524 public static final String TOPOLOGY_MESSAGE_TIMEOUT_SECS = "topology.message.timeout.secs";525 public static final Object TOPOLOGY_MESSAGE_TIMEOUT_SECS_SCHEMA = Number.class;526 527 /**528 * A list of serialization registrations for Kryo ( http://code.google.com/p/kryo/ ),529 * the underlying serialization framework for Storm. A serialization can either530 * be the name of a class (in which case Kryo will automatically create a serializer for the class531 * that saves all the object's fields), or an implementation of com.esotericsoftware.kryo.Serializer.532 *533 * See Kryo's documentation for more information about writing custom serializers.534 */535 public static final String TOPOLOGY_KRYO_REGISTER = "topology.kryo.register";536 public static final Object TOPOLOGY_KRYO_REGISTER_SCHEMA = ConfigValidation.StringsValidator;537 538 /**539 * A list of classes that customize storm's kryo instance during start-up.540 * Each listed class name must implement IKryoDecorator. During start-up the 541 * listed class is instantiated with 0 arguments, then its 'decorate' method 542 * is called with storm's kryo instance as the only argument.543 */544 public static final String TOPOLOGY_KRYO_DECORATORS = "topology.kryo.decorators";545 public static final Object TOPOLOGY_KRYO_DECORATORS_SCHEMA = ConfigValidation.StringsValidator;546 547 /**548 * Class that specifies how to create a Kryo instance for serialization. Storm will then apply549 * topology.kryo.register and topology.kryo.decorators on top of this. The default implementation550 * implements topology.fall.back.on.java.serialization and turns references off.551 */552 public static final String TOPOLOGY_KRYO_FACTORY = "topology.kryo.factory";553 public static final Object TOPOLOGY_KRYO_FACTORY_SCHEMA = String.class;554 555 556 /**557 * Whether or not Storm should skip the loading of kryo registrations for which it558 * does not know the class or have the serializer implementation. Otherwise, the task will559 * fail to load and will throw an error at runtime. The use case of this is if you want to560 * declare your serializations on the storm.yaml files on the cluster rather than every single561 * time you submit a topology. Different applications may use different serializations and so562 * a single application may not have the code for the other serializers used by other apps.563 * By setting this config to true, Storm will ignore that it doesn't have those other serializations564 * rather than throw an error.565 */566 public static final String TOPOLOGY_SKIP_MISSING_KRYO_REGISTRATIONS= "topology.skip.missing.kryo.registrations";567 public static final Object TOPOLOGY_SKIP_MISSING_KRYO_REGISTRATIONS_SCHEMA = Boolean.class;568 569 /*570 * A list of classes implementing IMetricsConsumer (See storm.yaml.example for exact config format).571 * Each listed class will be routed all the metrics data generated by the storm metrics API. 572 * Each listed class maps 1:1 to a system bolt named __metrics_ClassName#N, and it's parallelism is configurable.573 */574 public static final String TOPOLOGY_METRICS_CONSUMER_REGISTER = "topology.metrics.consumer.register";575 public static final Object TOPOLOGY_METRICS_CONSUMER_REGISTER_SCHEMA = ConfigValidation.MapsValidator;576 577 578 /**579 * The maximum parallelism allowed for a component in this topology. This configuration is580 * typically used in testing to limit the number of threads spawned in local mode.581 */582 public static final String TOPOLOGY_MAX_TASK_PARALLELISM="topology.max.task.parallelism";583 public static final Object TOPOLOGY_MAX_TASK_PARALLELISM_SCHEMA = Number.class;584 585 586 /**587 * The maximum number of tuples that can be pending on a spout task at any given time. 588 * This config applies to individual tasks, not to spouts or topologies as a whole. 589 * 590 * A pending tuple is one that has been emitted from a spout but has not been acked or failed yet.591 * Note that this config parameter has no effect for unreliable spouts that don't tag 592 * their tuples with a message id.593 */594 public static final String TOPOLOGY_MAX_SPOUT_PENDING="topology.max.spout.pending"; 595 public static final Object TOPOLOGY_MAX_SPOUT_PENDING_SCHEMA = Number.class;596 597 /**598 * A class that implements a strategy for what to do when a spout needs to wait. Waiting is599 * triggered in one of two conditions:600 * 601 * 1. nextTuple emits no tuples602 * 2. The spout has hit maxSpoutPending and can't emit any more tuples603 */604 public static final String TOPOLOGY_SPOUT_WAIT_STRATEGY="topology.spout.wait.strategy"; 605 public static final Object TOPOLOGY_SPOUT_WAIT_STRATEGY_SCHEMA = String.class;606 607 /**608 * The amount of milliseconds the SleepEmptyEmitStrategy should sleep for.609 */610 public static final String TOPOLOGY_SLEEP_SPOUT_WAIT_STRATEGY_TIME_MS="topology.sleep.spout.wait.strategy.time.ms";611 public static final Object TOPOLOGY_SLEEP_SPOUT_WAIT_STRATEGY_TIME_MS_SCHEMA = Number.class;612 613 /**614 * The maximum amount of time a component gives a source of state to synchronize before it requests615 * synchronization again.616 */617 public static final String TOPOLOGY_STATE_SYNCHRONIZATION_TIMEOUT_SECS="topology.state.synchronization.timeout.secs";618 public static final Object TOPOLOGY_STATE_SYNCHRONIZATION_TIMEOUT_SECS_SCHEMA = Number.class;619 620 /**621 * The percentage of tuples to sample to produce stats for a task.622 */623 public static final String TOPOLOGY_STATS_SAMPLE_RATE="topology.stats.sample.rate";624 public static final Object TOPOLOGY_STATS_SAMPLE_RATE_SCHEMA = Number.class;625 626 /**627 * The time period that builtin metrics data in bucketed into. 628 */629 public static final String TOPOLOGY_BUILTIN_METRICS_BUCKET_SIZE_SECS="topology.builtin.metrics.bucket.size.secs";630 public static final Object TOPOLOGY_BUILTIN_METRICS_BUCKET_SIZE_SECS_SCHEMA = Number.class;631 632 /**633 * Whether or not to use Java serialization in a topology.634 */635 public static final String TOPOLOGY_FALL_BACK_ON_JAVA_SERIALIZATION="topology.fall.back.on.java.serialization";636 public static final Object TOPOLOGY_FALL_BACK_ON_JAVA_SERIALIZATION_SCHEMA = Boolean.class;637 638 /**639 * Topology-specific options for the worker child process. This is used in addition to WORKER_CHILDOPTS.640 */641 public static final String TOPOLOGY_WORKER_CHILDOPTS="topology.worker.childopts";642 public static final Object TOPOLOGY_WORKER_CHILDOPTS_SCHEMA = String.class;643 644 /**645 * This config is available for TransactionalSpouts, and contains the id ( a String) for646 * the transactional topology. This id is used to store the state of the transactional647 * topology in Zookeeper.648 */649 public static final String TOPOLOGY_TRANSACTIONAL_ID="topology.transactional.id";650 public static final Object TOPOLOGY_TRANSACTIONAL_ID_SCHEMA = String.class;651 652 /**653 * A list of task hooks that are automatically added to every spout and bolt in the topology. An example654 * of when you'd do this is to add a hook that integrates with your internal 655 * monitoring system. These hooks are instantiated using the zero-arg constructor.656 */657 public static final String TOPOLOGY_AUTO_TASK_HOOKS="topology.auto.task.hooks";658 public static final Object TOPOLOGY_AUTO_TASK_HOOKS_SCHEMA = ConfigValidation.StringsValidator;659 660 661 /**662 * The size of the Disruptor receive queue for each executor. Must be a power of 2.663 */664 public static final String TOPOLOGY_EXECUTOR_RECEIVE_BUFFER_SIZE="topology.executor.receive.buffer.size";665 public static final Object TOPOLOGY_EXECUTOR_RECEIVE_BUFFER_SIZE_SCHEMA = ConfigValidation.PowerOf2Validator;666 667 /**668 * The maximum number of messages to batch from the thread receiving off the network to the 669 * executor queues. Must be a power of 2.670 */671 public static final String TOPOLOGY_RECEIVER_BUFFER_SIZE="topology.receiver.buffer.size";672 public static final Object TOPOLOGY_RECEIVER_BUFFER_SIZE_SCHEMA = ConfigValidation.PowerOf2Validator;673 674 /**675 * The size of the Disruptor send queue for each executor. Must be a power of 2.676 */677 public static final String TOPOLOGY_EXECUTOR_SEND_BUFFER_SIZE="topology.executor.send.buffer.size";678 public static final Object TOPOLOGY_EXECUTOR_SEND_BUFFER_SIZE_SCHEMA = ConfigValidation.PowerOf2Validator;679 680 /**681 * The size of the Disruptor transfer queue for each worker.682 */683 public static final String TOPOLOGY_TRANSFER_BUFFER_SIZE="topology.transfer.buffer.size";684 public static final Object TOPOLOGY_TRANSFER_BUFFER_SIZE_SCHEMA = Number.class;685 686 /**687 * How often a tick tuple from the "__system" component and "__tick" stream should be sent688 * to tasks. Meant to be used as a component-specific configuration.689 */690 public static final String TOPOLOGY_TICK_TUPLE_FREQ_SECS="topology.tick.tuple.freq.secs";691 public static final Object TOPOLOGY_TICK_TUPLE_FREQ_SECS_SCHEMA = Number.class;692 693 694 /**695 * Configure the wait strategy used for internal queuing. Can be used to tradeoff latency696 * vs. throughput697 */698 public static final String TOPOLOGY_DISRUPTOR_WAIT_STRATEGY="topology.disruptor.wait.strategy";699 public static final Object TOPOLOGY_DISRUPTOR_WAIT_STRATEGY_SCHEMA = String.class;700 701 /**702 * The size of the shared thread pool for worker tasks to make use of. The thread pool can be accessed 703 * via the TopologyContext.704 */705 public static final String TOPOLOGY_WORKER_SHARED_THREAD_POOL_SIZE="topology.worker.shared.thread.pool.size";706 public static final Object TOPOLOGY_WORKER_SHARED_THREAD_POOL_SIZE_SCHEMA = Number.class;707 708 /**709 * The interval in seconds to use for determining whether to throttle error reported to Zookeeper. For example, 710 * an interval of 10 seconds with topology.max.error.report.per.interval set to 5 will only allow 5 errors to be711 * reported to Zookeeper per task for every 10 second interval of time.712 */713 public static final String TOPOLOGY_ERROR_THROTTLE_INTERVAL_SECS="topology.error.throttle.interval.secs";714 public static final Object TOPOLOGY_ERROR_THROTTLE_INTERVAL_SECS_SCHEMA = Number.class;715 716 /**717 * See doc for TOPOLOGY_ERROR_THROTTLE_INTERVAL_SECS718 */719 public static final String TOPOLOGY_MAX_ERROR_REPORT_PER_INTERVAL="topology.max.error.report.per.interval";720 public static final Object TOPOLOGY_MAX_ERROR_REPORT_PER_INTERVAL_SCHEMA = Number.class;721 722 723 /**724 * How often a batch can be emitted in a Trident topology.725 */726 public static final String TOPOLOGY_TRIDENT_BATCH_EMIT_INTERVAL_MILLIS="topology.trident.batch.emit.interval.millis";727 public static final Object TOPOLOGY_TRIDENT_BATCH_EMIT_INTERVAL_MILLIS_SCHEMA = Number.class;728 729 /**730 * Name of the topology. This config is automatically set by Storm when the topology is submitted.731 */732 public static final String TOPOLOGY_NAME="topology.name";733 public static final Object TOPOLOGY_NAME_SCHEMA = String.class;734 735 /**736 * Max pending tuples in one ShellBolt737 */738 public static final String TOPOLOGY_SHELLBOLT_MAX_PENDING="topology.shellbolt.max.pending";739 public static final Object TOPOLOGY_SHELLBOLT_MAX_PENDING_SCHEMA = Number.class;740 741 /**742 * The root directory in ZooKeeper for metadata about TransactionalSpouts.743 */744 public static final String TRANSACTIONAL_ZOOKEEPER_ROOT="transactional.zookeeper.root";745 public static final Object TRANSACTIONAL_ZOOKEEPER_ROOT_SCHEMA = String.class;746 747 /**748 * The list of zookeeper servers in which to keep the transactional state. If null (which is default),749 * will use storm.zookeeper.servers750 */751 public static final String TRANSACTIONAL_ZOOKEEPER_SERVERS="transactional.zookeeper.servers";752 public static final Object TRANSACTIONAL_ZOOKEEPER_SERVERS_SCHEMA = ConfigValidation.StringsValidator;753 754 /**755 * The port to use to connect to the transactional zookeeper servers. If null (which is default),756 * will use storm.zookeeper.port757 */758 public static final String TRANSACTIONAL_ZOOKEEPER_PORT="transactional.zookeeper.port";759 public static final Object TRANSACTIONAL_ZOOKEEPER_PORT_SCHEMA = Number.class;760 761 /**762 * The number of threads that should be used by the zeromq context in each worker process.763 */764 public static final String ZMQ_THREADS = "zmq.threads";765 public static final Object ZMQ_THREADS_SCHEMA = Number.class;766 767 /**768 * How long a connection should retry sending messages to a target host when769 * the connection is closed. This is an advanced configuration and can almost770 * certainly be ignored.771 */772 public static final String ZMQ_LINGER_MILLIS = "zmq.linger.millis";773 public static final Object ZMQ_LINGER_MILLIS_SCHEMA = Number.class;774 775 /**776 * The high water for the ZeroMQ push sockets used for networking. Use this config to prevent buffer explosion777 * on the networking layer.778 */779 public static final String ZMQ_HWM = "zmq.hwm";780 public static final Object ZMQ_HWM_SCHEMA = Number.class;781 782 /**783 * This value is passed to spawned JVMs (e.g., Nimbus, Supervisor, and Workers)784 * for the java.library.path value. java.library.path tells the JVM where 785 * to look for native libraries. It is necessary to set this config correctly since786 * Storm uses the ZeroMQ and JZMQ native libs. 787 */788 public static final String JAVA_LIBRARY_PATH = "java.library.path";789 public static final Object JAVA_LIBRARY_PATH_SCHEMA = String.class;790 791 /**792 * The path to use as the zookeeper dir when running a zookeeper server via793 * "storm dev-zookeeper". This zookeeper instance is only intended for development;794 * it is not a production grade zookeeper setup.795 */796 public static final String DEV_ZOOKEEPER_PATH = "dev.zookeeper.path";797 public static final Object DEV_ZOOKEEPER_PATH_SCHEMA = String.class;798 799 /**800 * A map from topology name to the number of machines that should be dedicated for that topology. Set storm.scheduler801 * to backtype.storm.scheduler.IsolationScheduler to make use of the isolation scheduler.802 */803 public static final String ISOLATION_SCHEDULER_MACHINES = "isolation.scheduler.machines";804 public static final Object ISOLATION_SCHEDULER_MACHINES_SCHEMA = Map.class;805 806 public static void setDebug(Map conf, boolean isOn) {807 conf.put(Config.TOPOLOGY_DEBUG, isOn);808 } 809 810 public void setDebug(boolean isOn) {811 setDebug(this, isOn);812 }813 814 @Deprecated815 public void setOptimize(boolean isOn) {816 put(Config.TOPOLOGY_OPTIMIZE, isOn);817 } 818 819 public static void setNumWorkers(Map conf, int workers) {820 conf.put(Config.TOPOLOGY_WORKERS, workers);821 }822 823 public void setNumWorkers(int workers) {824 setNumWorkers(this, workers);825 }826 827 public static void setNumAckers(Map conf, int numExecutors) {828 conf.put(Config.TOPOLOGY_ACKER_EXECUTORS, numExecutors);829 }830 831 public void setNumAckers(int numExecutors) {832 setNumAckers(this, numExecutors);833 }834 835 public static void setMessageTimeoutSecs(Map conf, int secs) {836 conf.put(Config.TOPOLOGY_MESSAGE_TIMEOUT_SECS, secs);837 }838 839 public void setMessageTimeoutSecs(int secs) {840 setMessageTimeoutSecs(this, secs);841 }842 843 public static void registerSerialization(Map conf, Class klass) {844 getRegisteredSerializations(conf).add(klass.getName());845 }846 847 public void registerSerialization(Class klass) {848 registerSerialization(this, klass);849 }850 851 public static void registerSerialization(Map conf, Class klass, Class serializerClass) {852 Mapregister = new HashMap ();853 register.put(klass.getName(), serializerClass.getName());854 getRegisteredSerializations(conf).add(register); 855 }856 857 public void registerSerialization(Class klass, Class serializerClass) {858 registerSerialization(this, klass, serializerClass);859 }860 861 public void registerMetricsConsumer(Class klass, Object argument, long parallelismHint) {862 HashMap m = new HashMap();863 m.put("class", klass.getCanonicalName());864 m.put("parallelism.hint", parallelismHint);865 m.put("argument", argument);866 867 List l = (List)this.get(TOPOLOGY_METRICS_CONSUMER_REGISTER);868 if(l == null) { l = new ArrayList(); }869 l.add(m);870 this.put(TOPOLOGY_METRICS_CONSUMER_REGISTER, l);871 }872 873 public void registerMetricsConsumer(Class klass, long parallelismHint) {874 registerMetricsConsumer(klass, null, parallelismHint);875 }876 877 public void registerMetricsConsumer(Class klass) {878 registerMetricsConsumer(klass, null, 1L);879 }880 881 public static void registerDecorator(Map conf, Class klass) {882 getRegisteredDecorators(conf).add(klass.getName());883 }884 885 public void registerDecorator(Class klass) {886 registerDecorator(this, klass);887 }888 889 public static void setKryoFactory(Map conf, Class klass) {890 conf.put(Config.TOPOLOGY_KRYO_FACTORY, klass.getName());891 }892 893 public void setKryoFactory(Class klass) {894 setKryoFactory(this, klass);895 }896 897 public static void setSkipMissingKryoRegistrations(Map conf, boolean skip) {898 conf.put(Config.TOPOLOGY_SKIP_MISSING_KRYO_REGISTRATIONS, skip);899 }900 901 public void setSkipMissingKryoRegistrations(boolean skip) {902 setSkipMissingKryoRegistrations(this, skip);903 }904 905 public static void setMaxTaskParallelism(Map conf, int max) {906 conf.put(Config.TOPOLOGY_MAX_TASK_PARALLELISM, max);907 }908 909 public void setMaxTaskParallelism(int max) {910 setMaxTaskParallelism(this, max);911 }912 913 public static void setMaxSpoutPending(Map conf, int max) {914 conf.put(Config.TOPOLOGY_MAX_SPOUT_PENDING, max);915 }916 917 public void setMaxSpoutPending(int max) {918 setMaxSpoutPending(this, max);919 }920 921 public static void setStatsSampleRate(Map conf, double rate) {922 conf.put(Config.TOPOLOGY_STATS_SAMPLE_RATE, rate);923 } 924 925 public void setStatsSampleRate(double rate) {926 setStatsSampleRate(this, rate);927 } 928 929 public static void setFallBackOnJavaSerialization(Map conf, boolean fallback) {930 conf.put(Config.TOPOLOGY_FALL_BACK_ON_JAVA_SERIALIZATION, fallback);931 } 932 933 public void setFallBackOnJavaSerialization(boolean fallback) {934 setFallBackOnJavaSerialization(this, fallback);935 } 936 937 private static List getRegisteredSerializations(Map conf) {938 List ret;939 if(!conf.containsKey(Config.TOPOLOGY_KRYO_REGISTER)) {940 ret = new ArrayList();941 } else {942 ret = new ArrayList((List) conf.get(Config.TOPOLOGY_KRYO_REGISTER));943 }944 conf.put(Config.TOPOLOGY_KRYO_REGISTER, ret);945 return ret;946 }947 948 private static List getRegisteredDecorators(Map conf) {949 List ret;950 if(!conf.containsKey(Config.TOPOLOGY_KRYO_DECORATORS)) {951 ret = new ArrayList();952 } else {953 ret = new ArrayList((List) conf.get(Config.TOPOLOGY_KRYO_DECORATORS)); 954 }955 conf.put(Config.TOPOLOGY_KRYO_DECORATORS, ret);956 return ret;957 }958 }
转http://www.aboutyun.com/thread-8109-1-1.html