kafka-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Levani Kokhreidze (JIRA)" <j...@apache.org>
Subject [jira] [Created] (KAFKA-8727) Control over standby tasks host assignment
Date Mon, 29 Jul 2019 15:25:00 GMT
Levani Kokhreidze created KAFKA-8727:
----------------------------------------

             Summary: Control over standby tasks host assignment
                 Key: KAFKA-8727
                 URL: https://issues.apache.org/jira/browse/KAFKA-8727
             Project: Kafka
          Issue Type: New Feature
          Components: streams
            Reporter: Levani Kokhreidze


*Motivation*

As of now, Kafka Streams user has no control over to which host Kafka Streams application
will create standby task. In production deployments (especially in Kubernetes) it's quite
common to have multiple instances of the same Kafka Streams application deployed across more
than one "cluster" in order to have high availability of the system.

For example, if we have 6 Kafka Streams instances deployed across two clusters, we'll get
3 Kafka Streams instances per cluster. With the current implementation, Kafka Streams application
may create "standby task" in the same cluster as the active task which, is not the most optimal
solution, since, in case of cluster failure recovery time will be much bigger. This is especially
problematic for Kafka Streams application that manages large state.

 

*Possible Solution*

**It would be great if in the Kafka Streams configuration we could have a possibility to inject
dynamic environment variables and use that environment variables to control over where standby
task should be created.

For example, suppose I have active task *1_1* with environment variable: *CLUSTER_ID: main01* then
stnadby task for *1_1* should be created where *CLUSTER_ID* *!=* *main01*



--
This message was sent by Atlassian JIRA
(v7.6.14#76016)

Mime
View raw message