[ https://issues.apache.org/jira/browse/KAFKA-4084?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17030290#comment-17030290 ]
Harsha commented on KAFKA-4084: ------------------------------- [~junrao] I am not sure if we can avoid a broker overload problem even with replication throttle. For example, each broker in a 10 node cluster has around 4000 topic-partitions with 3 days of the retention period. If a broker goes down it needs to rebuild the data for the past 3 days for all of these 4000 partitions. If we apply a throttle of 200MB/sec for the new broker, its pulling from 9 other brokers each with 200MB/sec. If the auto.leader.rebalance.enable is enabled, this broker becomes a leader for topics as soon it catches up and be part of ISR. So it's still trying pull data from other brokers at 200MB/sec and while becoming a leader for few topic partitions which it will serve for both producer and consumer which increases the load on this broker. This may not be an issue in AWS with EBS volumes but it still causes an issue with on-prem hosts. > automated leader rebalance causes replication downtime for clusters with too > many partitions > -------------------------------------------------------------------------------------------- > > Key: KAFKA-4084 > URL: https://issues.apache.org/jira/browse/KAFKA-4084 > Project: Kafka > Issue Type: Bug > Components: controller > Affects Versions: 0.8.2.2, 0.9.0.0, 0.9.0.1, 0.10.0.0, 0.10.0.1 > Reporter: Tom Crayford > Priority: Major > Labels: reliability > Fix For: 1.1.0 > > > If you enable {{auto.leader.rebalance.enable}} (which is on by default), and > you have a cluster with many partitions, there is a severe amount of > replication downtime following a restart. This causes > `UnderReplicatedPartitions` to fire, and replication is paused. > This is because the current automated leader rebalance mechanism changes > leaders for *all* imbalanced partitions at once, instead of doing it > gradually. This effectively stops all replica fetchers in the cluster > (assuming there are enough imbalanced partitions), and restarts them. This > can take minutes on busy clusters, during which no replication is happening > and user data is at risk. Clients with {{acks=-1}} also see issues at this > time, because replication is effectively stalled. > To quote Todd Palino from the mailing list: > bq. There is an admin CLI command to trigger the preferred replica election > manually. There is also a broker configuration “auto.leader.rebalance.enable” > which you can set to have the broker automatically perform the PLE when > needed. DO NOT USE THIS OPTION. There are serious performance issues when > doing so, especially on larger clusters. It needs some development work that > has not been fully identified yet. > This setting is extremely useful for smaller clusters, but with high > partition counts causes the huge issues stated above. > One potential fix could be adding a new configuration for the number of > partitions to do automated leader rebalancing for at once, and *stop* once > that number of leader rebalances are in flight, until they're done. There may > be better mechanisms, and I'd love to hear if anybody has any ideas. -- This message was sent by Atlassian Jira (v8.3.4#803005)