You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@storm.apache.org by "Nick R. Katsipoulakis" <ni...@gmail.com> on 2015/06/24 20:14:38 UTC

Worker thread memory

Hello all,

I am working on an EC2 Storm cluster, and I want the workers in the
supervisor machines to use 4GBs of memory, so I add the following line in
the machine that hosts the nimbus:

worker.childopts-Xmx4096m -XX:+UseConcMarkSweepGC -XX:+UseParNewGC
-XX:+UseConcMarkSweepGC -XX:NewSize=128m
-XX:CMSInitiatingOccupancyFraction=70 -XX: -CMSConcurrentMTEnabled
Djava.net.preferIPv4Stack=true
However, when I take a look into the workers' logs (on each other machine
who is running a supervisor), I do not find the above line on the part that
launches the worker with the given arguments. In fact, I find the following
line:

2015-06-24T17:52:45.349+0000 b.s.d.worker [INFO] Launching worker for
tpch-q5-top-2-1435168361 on 5568726d-ad65-4a7c-ba52-32eed83276ad:6703 with
id 829f36fc-eeb9-4eef-ae89-9fb6565e9108 and conf {"dev.zookeeper.path"
"/tmp/dev-storm-zookeeper", "topology.tick.tuple.freq.secs" nil,
"topology.builtin.metrics.bucket.size.secs" 60,
"topology.fall.back.on.java.serialization" true,
"topology.max.error.report.per.interval" 5, "zmq.linger.millis" 5000,
"topology.skip.missing.kryo.registrations" false,
"storm.messaging.netty.client_worker_threads" 4, "ui.childopts" "-Xmx768m",
"storm.zookeeper.session.timeout" 20000, "nimbus.reassign" true,
"topology.trident.batch.emit.interval.millis" 500, "
storm.messaging.netty.flush.check.interval.ms" 10,
"nimbus.monitor.freq.secs" 10, "logviewer.childopts" "-Xmx128m",
"java.library.path" "/usr/local/lib:/opt/local/lib:/usr/lib", "storm.home"
"/opt/apache-storm-0.9.4", "topology.executor.send.buffer.size" 1024,
"storm.local.dir" "/mnt/storm", "storm.messaging.netty.buffer_size"
10485760, "supervisor.worker.start.timeout.secs" 120,
"topology.enable.message.timeouts" true, "nimbus.cleanup.inbox.freq.secs"
600, "nimbus.inbox.jar.expiration.secs" 3600, "drpc.worker.threads" 64,
"storm.meta.serialization.delegate"
"backtype.storm.serialization.DefaultSerializationDelegate",
"topology.worker.shared.thread.pool.size" 4, "nimbus.host" "52.25.74.163",
"storm.messaging.netty.min_wait_ms" 100, "storm.zookeeper.port" 2181,
"transactional.zookeeper.port" nil, "topology.executor.receive.buffer.size"
1024, "transactional.zookeeper.servers" nil, "storm.zookeeper.root"
"/storm", "storm.zookeeper.retry.intervalceiling.millis" 30000,
"supervisor.enable" true, "storm.messaging.netty.server_worker_threads" 4,
"storm.zookeeper.servers" ["172.31.28.73" "172.31.38.251" "172.31.38.252"],
"transactional.zookeeper.root" "/transactional", "topology.acker.executors"
nil, "topology.transfer.buffer.size" 1024, "topology.worker.childopts" nil,
"drpc.queue.size" 128, "worker.childopts" "-Xmx768m",
"supervisor.heartbeat.frequency.secs" 5,
"topology.error.throttle.interval.secs" 10, "zmq.hwm" 0, "drpc.port" 3772,
"supervisor.monitor.frequency.secs" 3, "drpc.childopts" "-Xmx768m",
"topology.receiver.buffer.size" 8, "task.heartbeat.frequency.secs" 3,
"topology.tasks" nil, "storm.messaging.netty.max_retries" 100,
"topology.spout.wait.strategy"
"backtype.storm.spout.SleepSpoutWaitStrategy",
"nimbus.thrift.max_buffer_size" 1048576, "topology.max.spout.pending" nil,
"storm.zookeeper.retry.interval" 1000, "
topology.sleep.spout.wait.strategy.time.ms" 1, "nimbus.topology.validator"
"backtype.storm.nimbus.DefaultTopologyValidator", "supervisor.slots.ports"
[6700 6701 6702 6703], "topology.environment" nil, "topology.debug" false,
"nimbus.task.launch.secs" 120, "nimbus.supervisor.timeout.secs" 60,
"topology.message.timeout.secs" 30, "task.refresh.poll.secs" 10,
"topology.workers" 1, "supervisor.childopts" "-Xmx256m",
"nimbus.thrift.port" 6627, "topology.stats.sample.rate" 0.05,
"worker.heartbeat.frequency.secs" 1, "topology.tuple.serializer"
"backtype.storm.serialization.types.ListDelegateSerializer",
"topology.disruptor.wait.strategy"
"com.lmax.disruptor.BlockingWaitStrategy", "topology.multilang.serializer"
"backtype.storm.multilang.JsonSerializer", "nimbus.task.timeout.secs" 30,
"storm.zookeeper.connection.timeout" 15000, "topology.kryo.factory"
"backtype.storm.serialization.DefaultKryoFactory", "drpc.invocations.port"
3773, "logviewer.port" 8000, "zmq.threads" 1, "storm.zookeeper.retry.times"
5, "topology.worker.receiver.thread.count" 1, "storm.thrift.transport"
"backtype.storm.security.auth.SimpleTransportPlugin",
"topology.state.synchronization.timeout.secs" 60,
"supervisor.worker.timeout.secs" 30, "nimbus.file.copy.expiration.secs"
600, "storm.messaging.transport" "backtype.storm.messaging.netty.Context", "
logviewer.appender.name" "A1", "storm.messaging.netty.max_wait_ms" 1000,
"drpc.request.timeout.secs" 600, "storm.local.mode.zmq" false, "ui.port"
8080, "nimbus.childopts" "-Xmx1024m", "storm.cluster.mode" "distributed",
"topology.max.task.parallelism" nil,
"storm.messaging.netty.transfer.batch.size" 262144, "topology.classpath"
nil}

which as you can see uses topology.worker.childopts: nil and
worker.childops: -Xmx768m. My question is the following: Do I need to add
the above line in the storm.yaml files of my supervisor nodes in order to
allow the JVM to use up to 4GBs of memory? Also, am I setting the right
value for what I am trying to achieve?

Thanks,
Nick