Child pages
  • KIP-122: Add Reset Consumer Group Offsets tooling
Skip to end of metadata
Go to start of metadata

Status

Current state: Under Discussion

Discussion thread: here 

JIRA: KAFKA-4743 

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

Motivation

This KIP wants to introduce a command-line tool to enable users to reset Consumer Group offsets.

Today, if users want to reprocess old records (i.e. records where offset is smaller than current consumer offset), they will need to modify client-side code (eg. using "KafkaConsumer#seek()" operations).

This process will involve write their own program which starts a consumer and commits offset or change client-side code, redeploy the application, and eventually rollback changes in application to consume records as usual. This is a cumbersome process given that users have to be aware of how many partitions are by topic, in which offset they are, and to which offset they want to move to.

Bringing the option to reset offsets from outside of the application will enable a cleaner and consistent way to achieve reprocessing and move along the topic.

Currently, there are 2 indexes that allow users to reprocess records: offset index and timestamp index.

Offset index represent a pointer over the sequential id that each record has by partition, available from Kafka inception.

Timestamp index (added in release 0.10.1.0 by KIP-33) represents a pointer over the epoch milliseconds from record creation, that could be defined by the Producer or the Broker, depending in the Timestamp Type.

From KIP-33:

"Searching by timestamp will have better accuracy. The guarantees provided are:

* The messages whose timestamp are after the searched timestamp will be consumed.

* Some messages with earlier timestamp might also be consumed."

 

Proposed Changes

We are considering to add the options in ConsumerGroupCommand to reset offsets.

There will be 3 execution options available:
  1. Plan: Print out the result of the operation (i.e. show a list of topic, partition, current offset, new offset to reset if the operation is executed). This will be the default option.
  2. Execute: Users will have to explicitly ask to execute the reset operation depending on the scenario and the scope specified.
  3. Export: Export plan to a CSV file to execute it later.

All these options are considered to be executed affecting only one Consumer Group.

The following scenarios will be supported:

IDScenarioDescription
1.Reset to DatetimeWhen we want to reset offsets to an specific point in time. (e.g. to 1/1/2017 at 00:00 to reprocess all records from these year)
2.Reset from DurationWhen we want to go back some period ago (e.g. P7D to reprocess all records from one week ago)
3.Reset to EarliestWhen we want to reprocess all the records available by partition.
4.Reset to LatestWhen a Consumer Groups has an offset lag and we don't want to process missing records, only move to the latest.
5.Reset to Current TimeWhen we want to only print out and/or backup the current offset by partition.
6.Reset to OffsetWhen we want to move to an specific offset.
7.Shift Offset by 'n'When we want to move forward or backward from the current position. 'n' can be a positive or negative value that will be add it to the current offset to move to a new position.
8.Reset from fileWhen we have a file with the required offsets by topic/partition to reset to.

 

And the following scopes:

  1. All Topics consumed by Consumer Group: This scope will consider all the topics that has been consumed by a Consumer Group.
  2. Specific List of Topics: This scope will consider all the topics defined by user.
  3. One Topic, All Partitions: This scope will consider only one topic and all partitions.
  4. One Topic, Specific Partition: This scope will consider only one topic and partition specified by user.

You can check the concept-proof implementation of this feature on this pull request.

Public Interfaces

Consumer Group Reset Offset options

Main option

--reset-offsets

This option should be executed independently from other Consumer Group options (list, describe, delete, etc.) and it will only support new consumers.

Required Arguments

IDArgumentTypeDescriptio
1.--groupRequiredConsumer Group ID.
2.--bootstrap-serverRequiredServer to connect to.

Scenarios

At least one of the scenarios should be defined to proceed with the execution

IDScenarioArgumentsConsiderationsExample
1.Reset to Datetime

--to-datetime YYYY-MM-DDTHH:mm:SS.sss±hh:mm

--to-datetime YYYY-MM-DDTHH:mm:SS.sssZ

--to-datetime YYYY-MM-DDTHH:mm:SS.sss

Datetime must be specified in ISO8601 format.

This option will translate the datetime to Epoch milliseconds, find the offsets by timestamp, and reset to those offsets.

If the Timezone is not specified, it will use UTC.

Reset to first offset since 01 January 2017, 00:00:00 hrs

--reset-offsets –group test.group --topic foo --to-datetime 2017-01-01T00:00:00Z

2.Reset by Duration--by-duration  PnDTnHnMnS

Duration must be specified in ISO8601 format.

This option will subtract the duration to the current timestamp in the server, and find the offsets using that subtracted timestamp, and reset to those offsets. The duration specified won't consider daylight saving effects.

Reset to first offset since one week ago (from current timestamp):

--reset-offsets --group test.group --topic foo --by-duration P7D

3.Reset to Earliest--to-earliestThis option will reset offsets to the earliest using Kafka Consumer's `#seekToBeginning`

Reset to earliest offset available:

--reset-offsets --group test.group --topic foo --to-earliest

4.Reset to Latest--to-latestThis option will reset offsets to the earliest using Kafka Consumer's `#seekToEnd`

Reset to latest offset available:

--reset-offsets --group test.group --topic foo --to-latest

5.Reset to Current Position(no scenario arguments)This option won't reset the offset. It will be used to print and export current offset.

Reset to current position:

--reset-offsets --group test.group --topic foo

6.Reset to Offset--to-offsetThis option will reset offsets to an specific value.

Reset to offset 1 in all partitions:

--reset-offsets --group test.group --topic foo --to-offset 1

7.Shift Offset by 'n'--shift-by n

This option will add the `n` value to the current offset, and reset to the result. `n` can be a positive or negative value, so offset will be move backward if it is negative, and forward if it is positive.

If current offset + n is higher than the latest offset, new offset will be set to latest.

If current offset + n is lower than the earliest offset, new offset will be set to earliest.

Reset to current offset plus 5 positions:

--reset-offsets --group test.group –topic foo --shift-by 5

8Reset from File--from-file PATH_TO_FILEThis option will take a Reset Plan CSV file with the offsets to reset by topics/partitions. It does not require scope, because topics and partitions are defined in the file.

Reset using a file with reset plan:

--reset-offsets --group test.group --from-file reset-plan.csv

Scopes

The scopes will be defined extending the existing `topic` argument, using the following format:

 

IDScopeArgumentsDescriptionExample
1.All topics--all-topicsIn this case the tool will run the scenario considering all topics that has been consumed by Consumer Group. It will consider all partitions by topic assigned.

Reset offsets in all topics and partitions:

--reset-offsets --group test.group --all-topics –to-earliest

2.List of Topics--topic topic1 --topic topic2In this case, the tool first will validate that input topics are been consumed by Consumer Group, and then run the scenario. It will consider all partitions by topic defined.

Reset offsets on all partitions of a list of topics:

--reset-offsets --group test.group --topic foo --topic bar –to-earliest

3.Topics and List of Partitions--topic topic1:0,1,2 --topic topic2:0In this case, the tool first will validate that input topic is been consumed by Consumer Group, and then run the scenario. It will consider only the partitions specified by topic.

Reset offsets on specific partitions by topics:

--reset-offsets --group test.group --topic foo:1,2 --topic bar:0,1,2,3 –to-earliest

4.Topic--topic topic1In this case, the tool first will validate that input topic is been consumed by Consumer Group, and then run the scenario. It will consider all partitions.

Reset offsets on a specifc topic, all partitions;

--reset-offsets --group test.group --topic foo --to-earliest

Execution Options

IDOptionArgumentsDescriptionExamples
1.Plan(no execution arguments)

This execution option will only print out the result of the scenario by scope.

The output will look like this:

TOPIC                 PARTITION NEW-OFFSET NEW-LAG LOG-END-OFFSET CONSUMER-ID HOST CLIENT-ID
foo 0 90 10 100 - - -

Prints result of resetting all topics and partitions to earliest:

--reset-offsets --group test.group --all-topics –to-earliest

2.Execute--executeThis execution option will run the reset offset process based on scenario and scope.

Prints and execute resetting all topics and partitions to earliest:

--reset-offsets --group test.group --all-topics –to-earliest --execute

3.Export--exportThis execution option will print out the reset plan in CSV format, that later could be used in the scenario 8. (i.e. as backup)

Prints plan for resetting all topics and partitions to earliest in CSV format:

--reset-offsets --group test.group --all-topics –to-earliest --export

Execute and Prints plan for resetting all topics and partitions to earliest in CSV format:

--reset-offsets --group test.group --all-topics –to-earliest --export --execute

 

Reset Plan File: CSV Format

The CSV schema will consist on:

<topic>,<partition>,<offset>

 

Sample:

topic-1,0,100
topic-1,1,200
topic-1,2,0
topic-2,0,0

Implementation Details 

These options will be implemented inside the `ConsumerGroupCommand`. The `resetOffsets` operation will looks like this:

(1) It will get assignments from `describeGroup` operation

(2) Then use `getAssignmentsToReset` to filter the assignments with the values defined by `–topic` of or `--all-topics`.

(3) Then get new offset by assignments using `prepareAssignmentsToReset`.

(4) It will be only executed when the ConsumerGroup selected is inactive o avoid race conditions. 

(5) It will be executed only if it is asked explictly. This will consist in create a Consumer using the same `group.id` and use:

To change the offsets.

 

Compatibility, Deprecation, and Migration Plan

This KIP won’t require a Migration Plan.


‘reset-to-datetime’ and ‘reset-to-period’ will require Timestamp Index, which will make them only available from version 0.10.1.0.


Nevertheless, it is possible to achieve the same result since release 0.10.0 using timestamp metadata by record, doing a linear search starting from latest. However, we are not considering to support this option because it will be an expensive/slow option.

All the other options will be available for releases from 0.10.0.0 given that KIP-97 is implemented.

This KIP will not support old consumers that store offsets in Zookeeper.

Test Plan

  • A unit test to validate that --reset-offset is executed independently from other Consumer Group options (list, delete, describe, etc.)
  • A unit test to validate that only one scenario is specified
  • A unit test to validate that only one scope is specified
  • A unit test to validate that only one execution option is specified
  • A unit test by combination of scenario, scope and execution option.
  • A unit test to validate that when calculated offset by partition is older that the earliest offset, tool resets offset to earliest (e.g. when we specified --reset-to 0, when earliest offset is 10)
  • A unit test to validate that when calculated offset by partition is bigger that the latest offset, tool resets offset to latest (e.g. when we specified --reset-to 0, when earliest offset is 10)

Rejected Alternatives

None

(no scenario arguments)

  • No labels