[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17201807#comment-17201807 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] looks the fix version is 2.7.0 &2.6.1, do u know when to release them? thanks. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.7.0, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17188870#comment-17188870 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] thanks, I've tested clients changes in PR #8834 against legacy Kafka server, it works fine as well. thanks for your response. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17188058#comment-17188058 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] one more question about PR #8834, whether or not *GroupCoordinator* changes is mandatory. I mean Kafka server changes should be more expensive that clients. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17187480#comment-17187480 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] I've tested against PR #8834, it works fine for our scenario. appreciate. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17185062#comment-17185062 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] cool, I'll try it later, thanks > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Comment Edited] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17180919#comment-17180919 ] Jerry Wei edited comment on KAFKA-10134 at 8/20/20, 8:53 AM: - [~guozhang] sure, I'll porting PR #9038 on 2.6.0 and share log with u. thx. please refer to the attached file: consumer3.log.2020-08-20.log was (Author: zhowei): [~guozhang] sure, I'll porting PR #9038 on 2.6.0 and share log with u. thx. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jerry Wei updated KAFKA-10134: -- Attachment: consumer3.log.2020-08-20.log > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer3.log.2020-08-20.log, > consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17180919#comment-17180919 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] sure, I'll porting PR #9038 on 2.6.0 and share log with u. thx. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.5.2, 2.6.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17177031#comment-17177031 ] Jerry Wei commented on KAFKA-10134: --- Kafka clients v2.6.0 is already released, but high cpu issue doesn't go away for our scenario and find there is a significant difference between v2.6.0 and PR #9011: [Kafka branch2.6|https://github.com/apache/kafka/blob/fe0279026bd3d854a7291f5bf99503469e900038/clients/src/main/java/org/apache/kafka/clients/consumer/KafkaConsumer.java#L1230] {code:java} if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer for join group updateAssignmentMetadataIfNeeded(timer, false); } else { ... } {code} [PR #9011|https://github.com/apache/kafka/pull/9011] {code:java} if (includeMetadataInTimeout) { // try to update assignment metadata; // do not need to block on the timer for join group if we have any fetchable partitions updateAssignmentMetadataIfNeeded(timer, !subscriptions.hasAnyFetchablePartitions()); } {code} > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17170601#comment-17170601 ] Jerry Wei commented on KAFKA-10134: --- The long time joining group is related to *max.poll.records*, looks like rejoin is trigged in *poll()*, it means only when processing complete and *poll()* be called. And *DisconnectException* should not be a real socket error, before leader consumer issues a rejoin request, looks like rejoin request from other consumers will be failed w/ *DisconnectException*. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17165187#comment-17165187 ] Jerry Wei commented on KAFKA-10134: --- [~ijuma] correct, just while restarting consumer, not broker restart. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17164290#comment-17164290 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] I'm using logback and PR: https://github.com/apache/kafka/pull/9011 included as well. yeah, just to restart consumer w/o broker change, both brokers were running on my local laptop, not found any other issue. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17162450#comment-17162450 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang] refer to the attached file (consumer5.log.2020-07-22.log), the last re-join happened at " 10:52:41.247 [pool-1-thread-1] INFO o.a.k.c.consumer.internals.AbstractCoordinator - [Consumer clientId=consumer5, groupId=consumerGroupId] (Re-)joining group" because there is one consumer trying to join. interesting, this time looks better than before, but still spend more than one min. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jerry Wei updated KAFKA-10134: -- Attachment: (was: consumer4.log.2020-07-22.log) > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jerry Wei updated KAFKA-10134: -- Attachment: consumer5.log.2020-07-22.log > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jerry Wei updated KAFKA-10134: -- Attachment: consumer4.log.2020-07-22.log > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > Attachments: consumer5.log.2020-07-22.log > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Comment Edited] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17157049#comment-17157049 ] Jerry Wei edited comment on KAFKA-10134 at 7/14/20, 12:21 AM: -- [~guozhang], yes, [~seanguo] and I are working on the same issue. I've verified PR:https://github.com/apache/kafka/pull/9011. looks like the CPU issue has gone away, but the new consumer spends much more long time joining group than before, it's about 2mins. was (Author: zhowei): [~guozhang], yes, [~seanguo] and I are working on the same issue. I've verified PR:https://github.com/apache/kafka/pull/9011. looks like the CPU issue is gone away, but the new consumer spends much more long time joining group than before, it's about 2mins. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17157049#comment-17157049 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang], yes, [~seanguo] and I are working on the same issue. I've verified PR:https://github.com/apache/kafka/pull/9011. looks like the CPU issue is gone away, but the new consumer spends much more long time joining group than before, it's about 2mins. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at > org.apache.kafka.clients.consumer.KafkaConsumer.updateAssignmentMetadataIfNeeded(KafkaConsumer.java:1275) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1241) > at > org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1216) > at > > By debugging into the code we found it looks like the clients are in a loop > on finding the coordinator. > I also tried the old rebalance protocol for the new version the issue still > exists but the CPU will be back to normal when the rebalance is done. > Also tried the same on the 2.4.1 which seems don't have this issue. So it > seems related something changed between 2.4.1 and 2.5.0. > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Comment Edited] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17156214#comment-17156214 ] Jerry Wei edited comment on KAFKA-10134 at 7/12/20, 6:29 AM: - [~guozhang], I have three brokers and 10 consumers. When I restart one of consumers, some of other consumers will be with high CPU issue. {code:java} // from KafkaConsumer.java (a fine fix) // private ConsumerRecords poll(final Timer timer, final boolean includeMetadataInTimeout); if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer if we still have // some assigned partitions, since even if we are 1) in the middle of a rebalance // or 2) have partitions with unknown starting positions we may still want to return some data // as long as there are some partitions fetchable; NOTE we always use a timer with 0ms // to never block on completing the rebalance procedure if there's any if (subscriptions.fetchablePartitions(tp -> true).isEmpty()) { updateAssignmentMetadataIfNeeded(timer); } else { final Timer updateMetadataTimer = time.timer(0L); updateAssignmentMetadataIfNeeded(updateMetadataTimer); timer.update(updateMetadataTimer.currentTimeMs()); } } else { while (!updateAssignmentMetadataIfNeeded(time.timer(Long.MAX_VALUE))) { log.warn("Still waiting for metadata"); } } {code} {code:java} // from KafkaConsumer.java (last commit) // private ConsumerRecords poll(final Timer timer, final boolean includeMetadataInTimeout); if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer for join group updateAssignmentMetadataIfNeeded(timer, false); } else { while (!updateAssignmentMetadataIfNeeded(time.timer(Long.MAX_VALUE), true)) { log.warn("Still waiting for metadata"); } } {code} Per the above two commits I have one question about `updateAssignmentMetadataIfNeeded(timer, false);` why the second parameter is false? Per my understanding when it's false, actually it's same as before `updateAssignmentMetadataIfNeeded(time.timer(0L));` I've tested w/ true, it looks fine for us and I'm thinking when it's w/ true, the behavior is similar to [the commit|https://github.com/apache/kafka/pull/8934/commits/333a967ec22ea22babf32b18349b76b6552a2fac]. was (Author: zhowei): [~guozhang], I have three brokers and 10 consumers. When I restart one of consumers, some of other consumers will be with high CPU issue. {code:java} // from KafkaConsumer.java (a fine fix) // private ConsumerRecords poll(final Timer timer, final boolean includeMetadataInTimeout); if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer if we still have // some assigned partitions, since even if we are 1) in the middle of a rebalance // or 2) have partitions with unknown starting positions we may still want to return some data // as long as there are some partitions fetchable; NOTE we always use a timer with 0ms // to never block on completing the rebalance procedure if there's any if (subscriptions.fetchablePartitions(tp -> true).isEmpty()) { updateAssignmentMetadataIfNeeded(timer); } else { final Timer updateMetadataTimer = time.timer(0L); updateAssignmentMetadataIfNeeded(updateMetadataTimer); timer.update(updateMetadataTimer.currentTimeMs()); } } else { while (!updateAssignmentMetadataIfNeeded(time.timer(Long.MAX_VALUE))) { log.warn("Still waiting for metadata"); } } {code} {code:java} // from KafkaConsumer.java (last commit) // private ConsumerRecords poll(final Timer timer, final boolean includeMetadataInTimeout); if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer for join group updateAssignmentMetadataIfNeeded(timer, false); } else { while (!updateAssignmentMetadataIfNeeded(time.timer(Long.MAX_VALUE), true)) { log.warn("Still waiting for
[jira] [Commented] (KAFKA-10134) High CPU issue during rebalance in Kafka consumer after upgrading to 2.5
[ https://issues.apache.org/jira/browse/KAFKA-10134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17156214#comment-17156214 ] Jerry Wei commented on KAFKA-10134: --- [~guozhang], I have three brokers and 10 consumers. When I restart one of consumers, some of other consumers will be with high CPU issue. {code:java} // from KafkaConsumer.java (a fine fix) // private ConsumerRecords poll(final Timer timer, final boolean includeMetadataInTimeout); if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer if we still have // some assigned partitions, since even if we are 1) in the middle of a rebalance // or 2) have partitions with unknown starting positions we may still want to return some data // as long as there are some partitions fetchable; NOTE we always use a timer with 0ms // to never block on completing the rebalance procedure if there's any if (subscriptions.fetchablePartitions(tp -> true).isEmpty()) { updateAssignmentMetadataIfNeeded(timer); } else { final Timer updateMetadataTimer = time.timer(0L); updateAssignmentMetadataIfNeeded(updateMetadataTimer); timer.update(updateMetadataTimer.currentTimeMs()); } } else { while (!updateAssignmentMetadataIfNeeded(time.timer(Long.MAX_VALUE))) { log.warn("Still waiting for metadata"); } } {code} {code:java} // from KafkaConsumer.java (last commit) // private ConsumerRecords poll(final Timer timer, final boolean includeMetadataInTimeout); if (includeMetadataInTimeout) { // try to update assignment metadata BUT do not need to block on the timer for join group updateAssignmentMetadataIfNeeded(timer, false); } else { while (!updateAssignmentMetadataIfNeeded(time.timer(Long.MAX_VALUE), true)) { log.warn("Still waiting for metadata"); } } {code} Per the above two commits I have one question about `updateAssignmentMetadataIfNeeded(timer, false);` why the second parameter is false? Per my understanding when it's false, actually it's same as before `updateAssignmentMetadataIfNeeded(time.timer(0L));` I've tested w/ true, it looks fine for us and I'm thinking when it's w/ true, the behavior is similar to [ the commit|https://github.com/apache/kafka/pull/8934/commits/333a967ec22ea22babf32b18349b76b6552a2fac]. > High CPU issue during rebalance in Kafka consumer after upgrading to 2.5 > > > Key: KAFKA-10134 > URL: https://issues.apache.org/jira/browse/KAFKA-10134 > Project: Kafka > Issue Type: Bug > Components: clients >Affects Versions: 2.5.0 >Reporter: Sean Guo >Assignee: Guozhang Wang >Priority: Blocker > Fix For: 2.6.0, 2.5.1 > > > We want to utilize the new rebalance protocol to mitigate the stop-the-world > effect during the rebalance as our tasks are long running task. > But after the upgrade when we try to kill an instance to let rebalance happen > when there is some load(some are long running tasks >30S) there, the CPU will > go sky-high. It reads ~700% in our metrics so there should be several threads > are in a tight loop. We have several consumer threads consuming from > different partitions during the rebalance. This is reproducible in both the > new CooperativeStickyAssignor and old eager rebalance rebalance protocol. The > difference is that with old eager rebalance rebalance protocol used the high > CPU usage will dropped after the rebalance done. But when using cooperative > one, it seems the consumers threads are stuck on something and couldn't > finish the rebalance so the high CPU usage won't drop until we stopped our > load. Also a small load without long running task also won't cause continuous > high CPU usage as the rebalance can finish in that case. > > "executor.kafka-consumer-executor-4" #124 daemon prio=5 os_prio=0 > cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 runnable > [0x7fe119aab000]"executor.kafka-consumer-executor-4" #124 daemon prio=5 > os_prio=0 cpu=76853.07ms elapsed=841.16s tid=0x7fe11f044000 nid=0x1f4 > runnable [0x7fe119aab000] java.lang.Thread.State: RUNNABLE at > org.apache.kafka.clients.consumer.internals.ConsumerCoordinator.poll(ConsumerCoordinator.java:467) > at >