Hello Sa Li,
Have you got your topology code uploaded in github? Maybe seeing the code
we can help you better.
Marcelo
2014-08-07 19:44 GMT+02:00 Sa Li sa.in.v...@gmail.com:
Hello, Kushan and Taylor
Thank you very much for the help, I exclude the zookeeper come with kafka
and storm-kafka
Hello, Kushan and Taylor
Thank you very much for the help, I exclude the zookeeper come with kafka and
storm-kafka packages in pom
!-- Kafka 0.8.0 compiled Scala 2.9.2 --
dependency
groupIdorg.apache.kafka/groupId
artifactIdkafka_2.10/artifactId
I see that your zookeeper is listening on port 2000. Is that how you have
configured the zookeeper?
--
Kushan Maskey
817.403.7500
On Tue, Aug 5, 2014 at 11:56 AM, Sa Li sa.in.v...@gmail.com wrote:
Thank you very much, Marcelo, it indeed worked, now I can run my code
without getting error.
Can you set zkPort in SpoutConfig to 2181 in your topology builder and see
if that helps?
--
Kushan Maskey
817.403.7500
On Wed, Aug 6, 2014 at 2:34 PM, Sa Li sa.in.v...@gmail.com wrote:
Hi, Kushan
You are completely right, I noticed this after you mentioned it,
apparently I am able to
You are running in local mode. So storm will start an in-process zookeeper for
it’s own use (usually on port 2000). In distributed mode, Storm will connect to
the zookeeper quorum specified in your storm.yaml.
In local mode, you would only need the external zookeeper for kafka and the
kafka
Thanks, Taylor, that makes sense, I check my kafka config, the
host.name=10.100.70.128,
and correspondingly change the spout config as
BrokerHosts zk = new ZkHosts(10.100.70.128);
TridentKafkaConfig spoutConf = new TridentKafkaConfig(zk, topictest);
it used to be localhost, actually
You have two different versions of zookeeper on the classpath (or in your
topology jar).
You need to find out where the conflicting zookeeper dependency is sneaking in
and exclude it.
If you are using maven 'mvn dependency:tree' and exclusions will help.
-Taylor
On Aug 6, 2014, at 6:14 PM,
Thank you very much, Marcelo, it indeed worked, now I can run my code without
getting error. However, another thing is keeping bother me, following is my
code:
public static class PrintStream implements Filter {
@SuppressWarnings(rawtypes”)
You are only sleeping for 100 milliseconds before shutting down the local
cluster, which is probably not long enough for the topology to come up and
start processing messages. Try increasing the sleep time to something like 10
seconds.
You can also reduce startup time with the following JVM
Thank you very much for your reply, Taylor. I tried to increase the sleep time
as 1 sec or 10 sec, however I got such error, it seems to be Async loop error.
Any idea about that?
3053 [Thread-19-$spoutcoord-spout0] INFO
org.apache.curator.framework.imps.CuratorFrameworkImpl - Starting
3058
If I set the sleep time as 1000 milisec, I got such error:
3067 [main] INFO backtype.storm.testing - Deleting temporary path
/tmp/0f1851f1-9499-48a5-817e-41712921d054
3163 [Thread-10-EventThread] INFO
com.netflix.curator.framework.state.ConnectionStateManager - State change:
SUSPENDED
3163
Can you let the topology run for 120 seconds or so? In my experience the kafka
bolt/spout takes a lot of latency initially as it tries to read/write from
zookeeper and initialize connections. On my mac it takes about 15 seconds
before the spout is actually opened.
Thanks
Parth
On Aug 5, 2014,
hello,
you can check your .jar application with command jar tf to see if class
kafka/api/OffsetRequest.class is part of the jar.
If not you can try to copy kafka-2.9.2-0.8.0.jar (or version you are using)
in storm_lib directory
Marcelo
2014-07-31 23:33 GMT+02:00 Sa Li sa.in.v...@gmail.com:
Hi, all
I am running a kafka-spout code in storm-server, the pom is
groupIdorg.apache.kafka/groupId
artifactIdkafka_2.9.2/artifactId
version0.8.0/version
scopeprovided/scope
exclusions
exclusion
14 matches
Mail list logo