THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
| Table of Contents |
|---|
Status
Current state: Under Discussion Accepted
Discussion thread: here [Change the link from the KIP proposal email archive to your own email thread]
[DISCUSS] KIP-379: Multiple Consumer Group Management
Voting thread: [VOTE] KIP-379: Multiple Consumer Group Management
JIRA: JIRA:
| Jira | ||||||
|---|---|---|---|---|---|---|
|
PULL REQUEST: https://github.com/apache/kafka/pull/5726/#5726
| Code Block | ||||
|---|---|---|---|---|
| ||||
Non-binding votes: 4 (Gwen Shapira, Ryanne Dolan, Jason Gustafson, Vahid Hashemian)
Binding votes: 1 (Damian Guy) |
Motivation
1) Running a set of commands like `--describe`, `--delete` or `--reset-offsets` for every single consumer group appears to be quite a tedios, time and memory consuming task. Each time you have to start a separate JVM to perform a query for a single group or spend some time writing your own client code.
2) Quite often one would like to perform operations on all consumer groups, thus, it would be great to have something as like an `--all-groups` option, similar to `--all-topics`.
3) In other cases, you would like to perform operations on a certain subset of consumer groups, thus, it would be great to have something as like a repeatable `--group` option for each group.
4) Performing a Describe query on all consumer groups with a combination of options `--describe` and `--state` will state` would allow us to grep consumer groups by their state (like `Stable, Dead, Empty, PreparingRebalance, NonExisting`). The only thing we have to do is to extend the "results table" with a GROUP column.
5) Support new New CSV file format to export consumer offsets for multiple groups.
Public Interfaces
1) Providing the command line tool kafka-consumer-groups.sh with support of repeating repeating --group parameter parameter for operations --describe, --delete, --reset-offsets:
...
3) Extending data output table format by adding "GROUP" column column for Describe queries:
--describe:
| GROUP | TOPIC | PARTITION | CURRENT-OFFSET | LOG-END-OFFSET | LAG | CONSUMER-ID | HOST | CLIENT-IDGROUP |
|---|
--describe --members:
| GROUP | CONSUMER-ID | HOST | CLIENT-ID | #PARTITIONS | GROUP |
|---|
--describe --members --verbose
| GROUP | CONSUMER-ID | HOST | CLIENT-ID | #PARTITIONSGROUP | ASSIGNMENT |
|---|
--describe --state
| GROUP | COORDINATOR (ID) | ASSIGNMENT-STRATEGY | STATE | #MEMBERSGROUP |
|---|
4) Update Add new CSV file to support multiple consumer groups. Thus, we will have to update table format by adding a consumer group column. . Backwards compatibility with old CSV file format should be preserved.
New CSV file format:
| GROUP | TOPIC | PARTITION | OFFSET |
|---|---|---|---|
Old CSV file format for a single group (leaving as is):
| TOPIC | PARTITION | OFFSET |
|---|---|---|
Proposed Changes
- Providing the command line tool kafka-consumer-groups.sh with support of repeating --group parameter for operations operations --describe, --delete, --reset-offsets
- Providing the command line tool kafka-consumer-groups.sh with an optional --all-groups parameter for operations operations --describe, --delete, --reset-offsets
- Extending data output table format by adding "GROUP" column for all all --describe queries queries
- Update CSV file to support multiple consumer groups.
- We will also have to make sure CSV generation is done properly since there are no restrictions on consumer group names and symbols like commas and quotes are allowed. That is, CSV file generation with export/import of consumer offsets needs some rework.
Compatibility, Deprecation, and Migration Plan
- Positive impact on existing users
- We are not changing behavior but adding new features
- Support backwards compatibility with old CSV file format for a single group case
Rejected Alternatives
None.
...