[ https://issues.apache.org/jira/browse/STORM-3531?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
khiemvk updated STORM-3531: --------------------------- Description: I deploy topology on cluster (3 Supervisors) with 3 workers, 3 Spouts read topic kafka with 3 partitions. Step 1: I kill worker1 on Supervisor1 Step 2: Worker1 automatic running again. Step 3: Kafka spout of worker1 register partition 1 and read not lag. Worker2 and worker3 error lag kafka, not continue read and commit offset Log worker: {color:#FF0000}Not polling on partition [TEST-0]. It has [250] uncommitted offsets, which exceeds the limit of [250]{color} was: I deploy topology on cluster (3 Supervisors) with 3 workers, 3 Spouts read topic kafka with 3 partitions. Step 1: I kill worker1 on Supervisor1 Step 2: Worker1 automatic running again. Step 3: Kafka spout of worker1 register partition 1 and read not lag. Worker2 and worker3 error lag kafka, not continue read and commit offset > Kafka Spouts Lag error when worker die > -------------------------------------- > > Key: STORM-3531 > URL: https://issues.apache.org/jira/browse/STORM-3531 > Project: Apache Storm > Issue Type: Bug > Components: storm-kafka-client > Affects Versions: 1.2.3 > Environment: Linux > Reporter: khiemvk > Priority: Major > > I deploy topology on cluster (3 Supervisors) with 3 workers, 3 Spouts read > topic kafka with 3 partitions. > Step 1: I kill worker1 on Supervisor1 > Step 2: Worker1 automatic running again. > Step 3: Kafka spout of worker1 register partition 1 and read not lag. Worker2 > and worker3 error lag kafka, not continue read and commit offset > Log worker: > {color:#FF0000}Not polling on partition [TEST-0]. It has [250] uncommitted > offsets, which exceeds the limit of [250]{color} -- This message was sent by Atlassian Jira (v8.3.4#803005)