Bryan, The server's shutdown hook should be able to trigger with any SIG except SIGKILL.
I just tried the following process and the stop script works for me: 1. bin/zookeeper-server-start.sh config/zookeeper.properties 2. bin/kafka-server-start.sh config/server.properties 3. bin/kafka-server-stop.sh ---- Kafka server logs ---- [2013-12-19 11:45:52,916] INFO Starting log flusher with a default period of 9223372036854775807 ms. (kafka.log.LogManager) [2013-12-19 11:45:52,960] INFO Awaiting socket connections on 0.0.0.0:9092. (kafka.network.Acceptor) [2013-12-19 11:45:52,961] INFO [Socket Server on Broker 0], Started (kafka.network.SocketServer) [2013-12-19 11:45:53,050] INFO Will not load MX4J, mx4j-tools.jar is not in the classpath (kafka.utils.Mx4jLoader$) [2013-12-19 11:45:53,070] INFO 0 successfully elected as leader (kafka.server.ZookeeperLeaderElector) [2013-12-19 11:45:53,192] INFO Registered broker 0 at path /brokers/ids/0 with address guwang-ld2.linkedin.biz:9092. (kafka.utils.ZkUtils$) [2013-12-19 11:45:53,215] INFO [Kafka Server 0], started (kafka.server.KafkaServer) [2013-12-19 11:45:53,303] INFO New leader is 0 (kafka.server.ZookeeperLeaderElector$LeaderChangeListener) [2013-12-19 11:45:53,381] INFO No state transitions triggered since no partitions are assigned to brokers 0 (kafka.utils.ZkUtils$) [2013-12-19 11:46:04,376] INFO [Kafka Server 0], shutting down (kafka.server.KafkaServer) [2013-12-19 11:46:04,380] INFO [Socket Server on Broker 0], Shutting down (kafka.network.SocketServer) [2013-12-19 11:46:04,386] INFO [Socket Server on Broker 0], Shutdown completed (kafka.network.SocketServer) [2013-12-19 11:46:04,387] INFO [Kafka Request Handler on Broker 0], shutting down (kafka.server.KafkaRequestHandlerPool) [2013-12-19 11:46:04,420] INFO [Kafka Request Handler on Broker 0], shut down completely (kafka.server.KafkaRequestHandlerPool) [2013-12-19 11:46:04,653] INFO [Replica Manager on Broker 0]: Shut down (kafka.server.ReplicaManager) [2013-12-19 11:46:04,654] INFO [ReplicaFetcherManager on broker 0] shutting down (kafka.server.ReplicaFetcherManager) [2013-12-19 11:46:04,655] INFO [ReplicaFetcherManager on broker 0] shutdown completed (kafka.server.ReplicaFetcherManager) [2013-12-19 11:46:04,661] INFO [Replica Manager on Broker 0]: Shutted down completely (kafka.server.ReplicaManager) [2013-12-19 11:46:04,673] INFO Terminate ZkClient event thread. (org.I0Itec.zkclient.ZkEventThread) [2013-12-19 11:46:04,684] INFO Session: 0x1430c64513b0000 closed (org.apache.zookeeper.ZooKeeper) [2013-12-19 11:46:04,684] INFO EventThread shut down (org.apache.zookeeper.ClientCnxn) [2013-12-19 11:46:04,685] INFO [Kafka Server 0], shut down completed (kafka.server.KafkaServer) Guozhang On Wed, Dec 18, 2013 at 9:01 PM, Jun Rao <jun...@gmail.com> wrote: > Maybe we should just change it to "kill -15". Could you try if that works? > Could you also file a jira? > > Thanks, > > Jun > > > On Tue, Dec 17, 2013 at 9:18 PM, Bryan Baugher <bjb...@gmail.com> wrote: > > > RHEL 6.4 64bit > > Java 6u35 > > > > > > On Tue, Dec 17, 2013 at 10:57 PM, Jun Rao <jun...@gmail.com> wrote: > > > > > Which OS are you on? > > > > > > Thanks, > > > > > > Jun > > > > > > > > > On Tue, Dec 17, 2013 at 11:15 AM, Bryan Baugher <bjb...@gmail.com> > > wrote: > > > > > > > Hi, > > > > > > > > We have been trying out the kafka 0.8.0 beta1 for awhile and recently > > > > attempted to upgrade to 0.8.0 but noticed that the stop server script > > > > doesn't seem to stop the broker anymore. I noticed here[1] that a > > commit > > > > was made before the release to change the signal sent to stop the > > broker > > > > from SIGTERM to SIGINT. Changing this script back to using SIGTERM > > seems > > > to > > > > fix the issue for me. Has anyone else noticed this, is there a bug or > > > > should I log one? > > > > > > > > [1] - > > > > > > > > > > > > > > https://github.com/apache/kafka/commit/51de7c55d2b3107b79953f401fc8c9530bd0eea0 > > > > > > > > -Bryan > > > > > > > > > > > > > > > -- > > -Bryan > > > -- -- Guozhang
