...

Currently, any broker node can be elected as the controller.  As part of this KIP, the active controller will instead be selected among a small potentially smaller pool of nodes specifically configured to act as controllers.  Typically three or five nodes in the cluster will be selected to be controllers.

...

Note that as long as at least one of the provided controller addresses is valid, the broker will be able to learn about the current metadata quorum and start up.  Once the broker is in contact with the metadata quorum, the quorum bootstrap addresses will not be needed.  This makes it possible to reconfigure the metadata quorum over time.  For example, if we start with a metadata quorum of host1, host2, host3, we could replace host3 with host4 without disrupting any of the brokers.  Then we could roll the brokers to apply the new metadata quorum bootstrap configuration of host1, host2, host4 on each one.

Node IDs

Just like brokers, controller nodes will have non-negative integer node IDs.  There will be a single ID space.  In other words, no controller should share the same ID as a broker.  Even when a broker and a controller are co-located in the same JVM, they must have different node IDs.

We define a node here as a tuple consisting of a node ID and a process role. Roles are defined by the `process.roles` configuration. As explained above, in a co-located configuration, a single process may take both the "controller" and "broker" roles. The node ID for both of these roles will be defined by the `node.id` configuration. However, this is mainly for configuration convenience. Semantically, we view the co-located process as representing two distinct nodes. Each node has its own listeners and its own set of APIs which it exposes. The APIs exposed by a controller node will not be the same as those exposed by a broker node.

Automatic node ID assignment via ZooKeeper will no longer be supported in KIP-500 mode.  Node IDs must be set in the configuration fileAutomatic node ID assignment via ZooKeeper will no longer be supported in KIP-500 mode.  Node IDs must be set in the configuration file for brokers and controllers.

Networking

Controller processes will listen on a separate endpoint from brokers.  This will be true even when the broker and controller are co-located in the same JVM.

In a well-run Kafka deployment, controller ports, like ZooKeeper ports, should be firewalled off from clients.  This will prevent clients from disrupting the cluster by flooding the controller ports with requests.  In the realm of ACLs, this translates to controllers requiring CLUSTERACTION on CLUSTER for all operations.  (KIP-590 describes how users' administrative requests will be forwarded to the controller quorum as needed.)

The only time when clients should contact a controller node directly is when they are debugging system issues.  This is similar to ZooKeeper, where we have things like zk-shell, but only for debugging.

Note that controllers do not appear in the MetadataResponses given to clients.

Metadata

The Metadata Topic

As described in KIP-500, the controller will store its data in the internal @metadata topic.__cluster_metadata topic.  This topic will contain a single partition which is managed by Raft, as described in KIP-595: A Raft Protocol for the Metadata Quorum.

...

Metadata changes need to be persisted to the @metadata __cluster_metadata log before we apply them on the other nodes in the cluster.  This means waiting for the metadata log's last stable offset to advance to the offset of the change.  After that point, we are guaranteed not to lose the change as long as we uphold the Raft invariants.

...

However, this "future state" may never be committed.  For example, the active controller might fail, truncating some of its future state.  Therefore, the active controller must not make this future state "visible" to the rest of the cluster until it has been made persistent – that is, until it becomes current state.  In the case of the @metadata topic__cluster_metadata topic, the replication protocol itself neatly takes care of this for us.  In the case of controller RPCs like AlterIsr, the controller handles this by not sending back a response until the designated change has been persisted.

...

The active controller makes changes to the metadata by appending records to the log. Each record has a null key, and this format for its value:

1. an unsigned varint specifying the frame type (currently 0)
2. an unsigned varint specifying the record type.
3. an unsigned varint specifying the record version
4. the payload in Kafka RPC format

For example, if we wanted to encode a TopicRecord, we might have 0 encoded as a varint, 1 encoded as a varint, followed by 0 as the record version, followed by the serialized topic data.

The frame type, record type, and version will typically only take one byte each, for a total header size overhead of two three bytes.

Record Format Versions

...

A broker only appears in MetadataResponse if it is in the "registered and active" state.  If it is in the unreigstered state, or the "registered and fenced" state, it will not appear in MetadataResponse.

Broker Registration

Every distributed system needs a way of managing cluster membership.  Prior to KIP-500, Kafka brokers registered ephemeral znodes in order to register themselves as part of the cluster.  The Kafka controller passively consumed the registration information from Zookeeper.

In the post-KIP-500 world there is no ZooKeeper and no ephemeral znodes.  Instead, each broker registers itself with the active controller using a BrokerRegistrationRequest. The active controller assigns the broker a new broker epoch, based on the next available offset in the log.  The new epoch is guaranteed to be higher than any previous epoch that has been used for the given broker id.

Each registration request contains a UUID which identifies the process which sent it.  This ID is called the incarnation ID.  This ensures that if the response to the registration request is lost, the broker can simply re-send the registration RPC and get the same successful result as before.

Registration requests also have information about the feature flags which the broker software supports.  The controller will refuse to register brokers if they don't support the feature flags which are active in the cluster.  In this case, the sysadmin needs to upgrade the broker software before it can be added to the cluster.

Handling Broker ID Conflicts

The controller only allows one broker process to be registered per broker ID.  Of course, broker processes go away occasionally-- for example, if a broker crashes.  A broker ID can be reused once a certain amount of time has gone past without any contact with the previous incarnation of the broker.

For the purpose of this section, handling a registration request or a broker heartbeat request are both considered forms of contact (even if the broker is fenced).

Broker Leases

Once a broker is registered, its next goal is to transition from registered+fenced to registered+active.  It can do this by sending a broker heartbeat request to the active controller.  When it receives a the heartbeat request, the active controller must decide whether to give the broker a lease, or whether it should deny the lease.  Brokers that don't have a broker ID lease are said to be "fenced."

As mentioned earlier, brokers which are fenced will not appear in MetadataResponses.  So clients that have up-to-date metadata will not try to contact fenced brokers.

Broker leases are time-bounded.  The amount of time they last is specified by the controller in its response.  Once the period has elapsed, if the broker has not renewed its lease via a heartbeat, it will be fenced.

Controlled Shutdown

In the pre-KIP-500 world, brokers triggered a controller shutdown by making an RPC to the controller.  When the controller returned a successful result from this RPC, the broker knew that it could shut down.

In the post-KIP-500 world, controller shutdown is handled by the broker heartbeat system instead.  In its periodic heartbeats, the broker asks the controller if it can transition into the SHUTDOWN state.  This motivates the controller to move all of the leaders off of that broker.  Once they are all moved, the controller responds to the heartbeat with a nextState of SHUTDOWN.

The Broker State Machine

NOT_RUNNING

This is the state that the broker is in before it has started up.  When the broker starts up, it transitions to STARTING.

STARTING

While the broker is in this state, it is trying to catch up with the latest metadata.  It fetches the metadata from the controller quorum. Once it has finished catching up, it transitions to the RECOVERY state.

RECOVERY

The broker is in this state while it is starting the log manager.  If the shutdown was clean, the broker will leave this state very quickly.  If the shutdown was unclean, the broker will stay in this state until log recovery is complete.

Once log recovery is done, the broker will start listening on the socket server.  It will then ask the controller to unfence it.  Once the controller agrees, it will transition to the RUNNING state.

RUNNING

The broker is in this state when it's up and running.

PENDING_CONTROLLED_SHUTDOWN

The broker is in this state when it has received a SIGTERM and is trying to shut down.

SHUTTING_DOWN

The broker is in this state when controlled shutdown has finished and it is shutting down.

Changes in the Broker State Machine

The numeric constants exposed through the metrics API have not changed, and there are no new or removed states.

The main change in the broker state machine is that the RECOVERING_FROM_UNCLEAN_SHUTDOWN state has been renamed to RECOVERY.  Also, unlike previously, the broker will always pass through RECOVERY (although it may only stay in this state for a very short amount of time).

Public Interfaces

meta.properties

When a storage directory is in use by a cluster running in kip-500 mode, it will have a new version of the meta.properties file.  Since the current version is 0, the new version will be 1.   Just as in version 0, meta.properties will continue to be a Java Properties file.  This essentially means that it is a plain text file where each line has the format key=value.

In version 0 of meta.properties, the cluster.id field is optional.  In contrast, in version 1 it is mandatory.

In version 0 of meta.properties, the cluster.id field is serialized in hexadecimal.  In contrast, in version 1 it is serialized in base64.

Version 1 of meta.properties adds the controller.id field.  Nodes which are acting as controllers will have this field.  If the node is also a broker, it will have the broker.id field as well.

During broker startup, if each directory does not have a broker.id, one will be added with the current broker id.  Similarly, during controller startup, if each directory does not have a controller id, one will be added with the current controller id.  If the broker id and/or controller id is present but does not match the configured id, the directory will be considered failed and will not be included in the valid storage directories of the process.

Here is an example of a version 1 meta.properties file:

$ cat /tmp/kafka-logs/meta.properties
#
#Tue Dec 01 10:08:08 PST 2020
cluster.id=3Db5QLSqSZieL3rJBUUegA
version=1
broker.id=0

Command-Line Tools

There will be a new command-line tool, kafka-storage.sh.

$ ./bin/kafka-storage.sh -h
usage: kafka-storage [-h] {info,format,random-uuid} ...

The Kafka storage tool.

positional arguments:
  {info,format,random-uuid}
    info                 Get information about the Kafka log directories on this node.
    format               Format the Kafka log directories on this node.
    random-uuid          Print a random UUID.

optional arguments:
  -h, --help             show this help message and exit

kafka-storage.sh will have three subcommands: info, format, and random-uuid.

info

$ ./bin/kafka-storage.sh info -h
usage: kafka-storage info [-h] --config CONFIG

optional arguments:
  -h, --help             show this help message and exit
  --config CONFIG, -c CONFIG
                         The Kafka configuration file to use.

The info command will give information about the configured storage directories.  Example output:

Found log directory:
  /tmp/kafka-logs

Found metadata: MetaProperties(clusterId=51380268-1036-410d-a8fc-fb3b55f48033)

format

$ ./bin/kafka-storage.sh format -h
usage: kafka-storage format [-h] --config CONFIG --cluster-id CLUSTER_ID [--ignore-formatted]

optional arguments:
  -h, --help             show this help message and exit
  --config CONFIG, -c CONFIG
                         The Kafka configuration file to use.
  --cluster-id CLUSTER_ID, -t CLUSTER_ID
                         The cluster ID to use.
  --ignore-formatted, -g

When running kip-500 mode, the storage directories must be formatted using this command prior to starting up the brokers and controllers.

If any of the storage directories are formatted, the command will normally fail.  This behavior can be changed by passing the --ignore-formatted option.  When this option is passed, the format command will skip over already formatted directories rather than failing.

random-uuid

$ ./bin/kafka-storage.sh random-uuid -h
usage: kafka-storage random-uuid [-h]

optional arguments:
  -h, --help             show this help message and exit

The random-uuid command prints out a random UUID to stdout.

$ ./bin/kafka-storage.sh random-uuid
51380268-1036-410d-a8fc-fb3b55f48033

Configurations

...

null

broker

controller

broker,controller

...

If this is null (absent) then we are in legacy mode.

Otherwise, we are in KIP-500 mode and this configuration determines what roles this process should play: broker, controller, or both.

...

If non-null, this must be a comma-separated list of listener names.

When communicating with the controller quorum, the broker will always use the first listener in this list.

...

A comma-separated list of the names of the listeners used by the KIP-500 controller. This configuration is required if this process is a KIP-500 controller. The legacy controller will not use this configuration

Despite the similar name, note that this is different from the "control plane listener" introduced by KIP-291.

...

A comma-separated list of the configured listeners.  For example,

INTERNAL://192.1.1.8:9092, EXTERNAL://10.1.1.5:9093, CONTROLLER://192.1.1.8:9094

...

If non-null, this must be a comma-separated list of all the controller voters, in the format:

{controller-id}@{controller-host):{controller-port}

...

When in KIP-500 mode, each node must have this configuration, in order to find out how to communicate with the controller quorum.

Note that this replaces the "quorum.voters" config described in KIP-595.

This configuration is required for both brokers and controllers.

...

a 32-bit ID

...

The controller id for this server.  Only required if this server is a controller.

...

controller.quorum.election.timeout.ms

...

Essentially, registration relates to the permanent state of the cluster: the number of brokers we expect to have, what racks we expect them to be in, and so on.  Fencing refers to the transitory state of the cluster: which brokers are currently available.

By separating the permanent state from the transitory state, we can more effectively handle transitory issues.  For example, if you have a 3 node cluster that is undergoing rolling upgrade, one of the nodes might be down because it is rolling.  However, we should still allow users to create new topics with replication factor 3. Currently, that is not possible, because the node's registration information gets wiped the moment its ZK registration goes away.  With KIP-631, the registration remains, although the node becomes fenced.  Another example is doing reassignment on a cluster where one or more nodes is down.  Currently, when a node is down, all of its ZK registration information is gone.  But  we need this information in order to understand things like whether the replicas of a particular partition are well-balanced across racks.

Broker Registration

Every distributed system needs a way of managing cluster membership.  Prior to KIP-500, Kafka brokers registered ephemeral znodes in order to register themselves as part of the cluster.  The Kafka controller passively consumed the registration information from Zookeeper.

In the post-KIP-500 world there is no ZooKeeper and no ephemeral znodes.  Instead, each broker registers itself with the active controller using a BrokerRegistrationRequest. The active controller assigns the broker a new broker epoch, based on the next available offset in the log.  The new epoch is guaranteed to be higher than any previous epoch that has been used for the given broker id.

Each registration request contains a UUID which identifies the process which sent it.  This ID is called the incarnation ID.  This ensures that if the response to the registration request is lost, the broker can simply re-send the registration RPC and get the same successful result as before.

Registration requests also have information about the feature flags which the broker software supports.  The controller will refuse to register brokers if they don't support the feature flags which are active in the cluster.  In this case, the sysadmin needs to upgrade the broker software before it can be added to the cluster.

Handling Broker ID Conflicts

The controller only allows one broker process to be registered per broker ID.  Of course, broker processes go away occasionally-- for example, if a broker crashes.  A broker ID can be reused once a certain amount of time has gone past without any contact with the previous incarnation of the broker.

For the purpose of this section, handling a registration request or a broker heartbeat request are both considered forms of contact (even if the broker is fenced).

Broker Leases

When the broker first starts up, it doesn't want to be unfenced immediately.  The reason is because it needs time to perform log recovery and some other startup tasks.  It is good for the broker to be fenced during this time, so that clients do not try to contact it and fail.  The broker indicates that it is not ready to be unfenced by setting ShouldFence = true in the heartbeats it sends out during this period.

Once the broker is ready to be unfenced, it starts setting ShouldFence = false in the heartbeats it sends out.  This makes it eligible for unfencing.  However, the controller will not actually unfence the broker unless its metadata is reasonably current.  The controller determines this by examining the metadata offset in the heartbeat request.

As mentioned earlier, brokers which are fenced will not appear in MetadataResponses.  So clients that have up-to-date metadata will not try to contact fenced brokers.

Broker leases are time-bounded.  Once the period has elapsed, if the broker has not renewed its lease via a heartbeat, it will be fenced.

Controlled Shutdown

In the pre-KIP-500 world, brokers triggered a controller shutdown by making an RPC to the controller.  When the controller returned a successful result from this RPC, the broker knew that it could shut down.

In the post-KIP-500 world, controller shutdown is handled by the broker heartbeat system instead.  In its periodic heartbeats, the broker asks the controller if it can transition into the controlled shutdown state.  It does this by setting the WantShutDown boolean.  This motivates the controller to move all of the leaders off of that broker.  Once they are all moved, the controller responds to the heartbeat with ShouldShutDown = true.  At that point, the broker knows it's safe to begin the shutdown process proper.

The Broker State Machine

NOT_RUNNING

This is the state that the broker is in before it has started up.  When the broker starts up, it transitions to STARTING.

STARTING

While the broker is in this state, it is trying to catch up with the latest metadata.  It fetches the metadata from the controller quorum. Once it has finished catching up, it transitions to the RECOVERY state.

RECOVERY

The broker is in this state while it is starting the log manager.  If the shutdown was clean, the broker will leave this state very quickly.  If the shutdown was unclean, the broker will stay in this state until log recovery is complete.

Once log recovery is done, the broker will start listening on the socket server.  It will then ask the controller to unfence it.  Once the controller agrees, it will transition to the RUNNING state.

RUNNING

The broker is in this state when it's up and running.

PENDING_CONTROLLED_SHUTDOWN

The broker is in this state when it has received a SIGTERM and is trying to shut down.

SHUTTING_DOWN

The broker is in this state when controlled shutdown has finished and it is shutting down.

Changes in the Broker State Machine

The numeric constants exposed through the metrics API have not changed, and there are no new or removed states.

The main change in the broker state machine is that the RECOVERING_FROM_UNCLEAN_SHUTDOWN state has been renamed to RECOVERY.  Also, unlike previously, the broker will always pass through RECOVERY (although it may only stay in this state for a very short amount of time).

Partition Reassignment

Just like the ZooKeeper-based controller, the quorum controller will implement the KIP-455 partition reassignment API. This API specifies both alter and list operations.

The alter operation specifies a partition, plus a target replica list. If the target replica list is a subset of the partition's current replica list, and the new replica list contains at least one member of the current ISR, then the reassignment can be completed immediately. In this case, the controller issues a PartitionChangeRecord changing the replica set and ISR (if appropriate).

On the other hand, if the new replica list has additional replicas, or specifies a subset of replicas that doesn't intersect with the current ISR, then we add the additional replicas to the replica list, and wait for the ISR to catch up.

So, for example, if the old replica list was [1, 2, 3], and the new list was [3, 4, 5], we would proceed as follows:

• change the replica list to [1, 2, 3, 4, 5], and set addingReplicas to [4, 5]
• Wait for an alterIsr which adds both broker 4 and 5
• Change the replica list to [3, 4, 5 ] and prune the ISR accordingly. Remove addingReplicas.

Just like with the ZooKeeper-based controller, the new controller stores reassignments as "hard state." Reassignments will continue even after a controller failover or broker shutdown and restart. Reassignment state appears in controller snapshots, as addingReplicas and removingReplicas.

Unlike the old controller, the quorum controller does not prevent topics that are undergoing reassignment from being deleted. If a topic is deleted, then any partition reassignments that it had are terminated.

Typically, reassignments which require ISR changes are completed by a leader adding some new replicas, via the alterIsr RPC. (Note that another possible way for a reassignment to be completed is via a broker being added to a partition ISR during unfencing). 

When a broker makes an alterIsr change that completes a reassignment, the resulting ISR will be different than requested. For example, if the leader is 1, and the current ISR is 1, 2, 3, and the target replica set is 1, 2, 4, when the leader tries to change the ISR to 1, 2, 3, 4, the controller will change it to 1, 2, 4 instead, completing the reassignment. Since the alterIsr response returns the actual ISR which was actually applied, the leader will apply this new ISR instead.

If the new ISR would does not contain the leader which made the alterIsr call, the controller returns FENCED_LEADER_EPOCH. This will notify the broker that it should wait until it can replay the latest partition change from the log. That partition change record will make this broker a follower rather than a leader.

Public Interfaces

meta.properties

When a storage directory is in use by a cluster running in kip-500 mode, it will have a new version of the meta.properties file.  Since the current version is 0, the new version will be 1.   Just as in version 0, meta.properties will continue to be a Java Properties file.  This essentially means that it is a plain text file where each line has the format key=value.

In version 0 of meta.properties, the cluster.id field is optional.  In contrast, in version 1 it is mandatory.

In version 0 of meta.properties, the cluster.id field is serialized in hexadecimal.  In contrast, in version 1 it is serialized in base64.

Version 1 of meta.properties replaces the broker.id field with node.id.

For servers running in kip-500 mode, the `meta.properties` file must be present in every log directory. The process will raise an error during startup if if either the meta.properties file does not exist or if the node.id found does not match what the value from the configuration file.

Here is an example of a version 1 meta.properties file:

$ cat /tmp/kafka-logs/meta.properties
#
#Tue Dec 01 10:08:08 PST 2020
cluster.id=3Db5QLSqSZieL3rJBUUegA
version=1
node.id=0

kafka-storage.sh

There will be a new command-line tool, kafka-storage.sh.

$ ./bin/kafka-storage.sh -h
usage: kafka-storage [-h] {info,format,random-uuid} ...

The Kafka storage tool.

positional arguments:
  {info,format,random-uuid}
    info                 Get information about the Kafka log directories on this node.
    format               Format the Kafka log directories on this node.
    random-uuid          Print a random UUID.

optional arguments:
  -h, --help             show this help message and exit

kafka-storage.sh will have three subcommands: info, format, and random-uuid.

info

$ ./bin/kafka-storage.sh info -h
usage: kafka-storage info [-h] --config CONFIG

optional arguments:
  -h, --help             show this help message and exit
  --config CONFIG, -c CONFIG
                         The Kafka configuration file to use.

The info command will give information about the configured storage directories.  Example output:

Found log directory:
  /tmp/kafka-logs

Found metadata: MetaProperties(clusterId=51380268-1036-410d-a8fc-fb3b55f48033)

format

$ ./bin/kafka-storage.sh format -h
usage: kafka-storage format [-h] --config CONFIG --cluster-id CLUSTER_ID [--ignore-formatted]

optional arguments:
  -h, --help             show this help message and exit
  --config CONFIG, -c CONFIG
                         The Kafka configuration file to use.
  --cluster-id CLUSTER_ID, -t CLUSTER_ID
                         The cluster ID to use.
  --ignore-formatted, -g

When running kip-500 mode, the storage directories must be formatted using this command prior to starting up the brokers and controllers.

If any of the storage directories are formatted, the command will normally fail.  This behavior can be changed by passing the --ignore-formatted option.  When this option is passed, the format command will skip over already formatted directories rather than failing.

random-uuid

$ ./bin/kafka-storage.sh random-uuid -h
usage: kafka-storage random-uuid [-h]

optional arguments:
  -h, --help             show this help message and exit

The random-uuid command prints out a random UUID to stdout.

$ ./bin/kafka-storage.sh random-uuid
8XUwXa9qSyi9tSOquGtauQ

kafka-cluster.sh

There will also be new command-line tool named kafka-cluster.sh.

$ ./bin/kafka-cluster.sh -h
usage: kafka-cluster [-h] {id,unregister} ...

The Kafka cluster tool.

positional arguments:
  {cluster-id,unregister}
    cluster-id           Get information about the ID of a cluster.
    unregister           Unregister a broker ID.

optional arguments:
  -h, --help             show this help message and exit

kafka-storage.sh will have two subcommands: id and unregister.

cluster-id

$ ./bin/kafka-cluster.sh cluster-id -h
usage: kafka-cluster cluster-id [-b,-c,-h]

optional arguments:
  -b, --bootstrap-server a list of host/port pairs to use for establishing the connection to the kafka cluster.
  -c, --config           a property file containing configs to be passed to Admin Client.
  -h, --help             show this help message and exit

The ID command prints out the cluster id

unregister

$ ./bin/kafka-cluster.sh unregister -h
usage: kafka-cluster unregister [-h] [--bootstrap-server BOOTSTRAP_SERVER] [--config CONFIG] [--id ID]

optional arguments:
  -h, --help             show this help message and exit
  --bootstrap-server BOOTSTRAP_SERVER, -b BOOTSTRAP_SERVER
                         A list of host/port pairs to use for establishing the connection to the kafka cluster.
  --config CONFIG, -c CONFIG
                         A property file containing configs to passed to AdminClient.
  --id ID, -i ID         The ID of the broker to unregister.

The decommission command removes the registration of a specific broker ID.  It will use make an UnregisterBrokerRequest in order to do this.

Changes to kafka-dump-log-segments.sh

kafka-dump-log-seguments.sh will have two new flags: --cluster-metadata-decoder, and --skip-record-metadata.

The --cluster-metadata-decoder flag will tell the DumpLogSegments tool to decode the records as KIP-500 metadata.  Each record will be output in the following JSON format:

payload: {"type":<record type enum name>, "version":<record version number>, "data":<record JSON>}

Example output:

 payload: {"type":"TOPIC_RECORD","version":0,"data":{"name":"bar","topicId":"GU_rXds2FGppL1JqXYpx2g"}}
 payload: {"type":"PARTITION_RECORD","version":0,"data":{"partitionId":0,"topicId":"GU_rXds2FGppL1JqXYpx2g","replicas":[1],"isr":[1],"removingReplicas":null,"addingReplicas":null,"leader":1,"leaderEpoch":0,"partitionEpoch":0}}
 payload: {"type":"PARTITION_CHANGE_RECORD","version":0,"data":{"partitionId":0,"topicId":"GU_rXds2FGppL1JqXYpx2g","leader":-1}}
 payload: {"type":"PARTITION_CHANGE_RECORD","version":0,"data":{"partitionId":0,"topicId":"WCnrza5uWKeerYa7HCNpOg","leader":-1}}
 payload: {"type":"PARTITION_CHANGE_RECORD","version":0,"data":{"partitionId":0,"topicId":"GU_rXds2FGppL1JqXYpx2g","leader":-1}}
 payload: {"type":"PARTITION_CHANGE_RECORD","version":0,"data":{"partitionId":0,"topicId":"WCnrza5uWKeerYa7HCNpOg","leader":-1}}
 payload: {"type":"FENCE_BROKER_RECORD","version":0,"data":{"id":1,"epoch":0}}

The --skip-record-metadata flag will skip printing metadata for each record.  However, metadata for each record batch will still be printed when this flag is present.

kafka-metadata-shell.sh

The Kafka Metadata shell is a new command which allows users to interactively examine the metadata stored in a KIP-500 cluster.

It can read the metadata from the controllers directly, by connecting to them, or from a metadata snapshot on disk.  In the former case, the quorum voters must be specified by passing the --controllers flag; in the latter case, the snapshot file should be specified via --snapshot.

$ ./bin/kafka-metadata-shell.sh -h
usage: metadata-tool [-h] [--controllers CONTROLLERS] [--config CONFIG] [--snapshot SNAPSHOT] [command [command ...]]

The Apache Kafka metadata tool

positional arguments:
  command                The command to run.

optional arguments:
  -h, --help             show this help message and exit
  --controllers CONTROLLERS, -C CONTROLLERS
                         The quorum voter connection string to use.
  --config CONFIG, -c CONFIG
                         The configuration file to use.
  --snapshot SNAPSHOT, -s SNAPSHOT
                         The snapshot file to read.

The metadata tool works by replaying the log and storing the state into in-memory nodes.  These nodes are presented in a fashion similar to filesystem directories.  For browsing the nodes, several commands are supported:

>> help
Welcome to the Apache Kafka metadata shell.

usage:  {cat,cd,exit,find,help,history,ls,man,pwd} ...

positional arguments:
  {cat,cd,exit,find,help,history,ls,man,pwd}
    cat                  Show the contents of metadata nodes.
    cd                   Set the current working directory.
    exit                 Exit the metadata shell.
    find                 Search for nodes in the directory hierarchy.
    help                 Display this help message.
    history              Print command history.
    ls                   List metadata nodes.
    man                  Show the help text for a specific command.
    pwd                  Print the current working directory.

The interface of the metadata tool is currently considered unstable and may change when KIP-500 becomes production-ready.

Configurations

Deprecated Configurations

New Error Codes

DUPLICATE_BROKER_REGISTRATION

There will be a new error code, DUPLICATE_BROKER_REGISTRATION, that the active controller will return when a broker tries to register with an ID that is currently in use.

INVALID_CLUSTER_ID

There will be a new error code, INVALID_CLUSTER_ID, that the controller will return if the broker tries to register with the wrong cluster ID.

AdminClient

There will be a new AdminClient RPC, unregisterBroker.

UnregisterBrokerResult unregisterBroker(int brokerId, UnregisterBrokerOptions options)

public class UnregisterBrokerResult {
    public KafkaFuture<Void> all();
}

public class UnregisterBrokerOptions extends AbstractOptions<UnregisterBrokerOptions> {
}

RPCs

Obsoleting the Metadata Propagation RPCs

As discussed earlier, the new controller will use FetchRequest to fetch metadata from the active controller.  The details of how Raft fetching will work are spelled out in KIP-595: A Raft Protocol for the Metadata Quorum.

Since we propagate the metadata via Raft, we will no longer need to send out LeaderAndIsrRequest, UpdateMetadataRequest, and StopReplicaRequest.  These requests will be sent out only when we're in legacy mode, not when we're in KIP-500 mode.  Eventually we will add some support for these requests to the new controller, in order to support rolling upgrade from a pre-KIP-500 release. However, for the purpose of this KIP, the new controller will not use these requests.

Obsoleting the Controlled Shutdown RPC

The broker heartbeat mechanism replaces the controlled shutdown RPC.  Therefore, we will not need to support the this RPC any more in the controller-- except for compatibility during upgrades, which will be described further in a follow-on KIP.

Topic Identifiers

This KIP builds on top of the work done in KIP-516 to improve how partitions are tracked.

Because each topic is identified by a unique topic UUID, we can implement topic deletion with a single record, RemoveTopicRecord.  Upon replaying this record, each broker will delete the associated topic if it is present.

Of course, some brokers may be down when the topic is deleted.  In fact, some brokers may never see the RemoveTopicRecord.  This record may get collapsed into one of the periodic metadata snapshots.  If this happens, the record will be reflected in the snapshot through the absence of a broker record, not its presence.  Therefore, during the startup process, brokers must compare the log directories that they have with the ones contained in the latest metadata.  The appropriate time to do this is at the start of the RECOVERY phase.  At this point, the broker has the latest metadata.

BrokerRegistration

Required ACLs: CLUSTERACTION on CLUSTER

{
  "apiKey": 57,
  "type": "request",
  "name": "BrokerRegistrationRequest",
  "validVersions": "0",
  "flexibleVersions": "0+",
  "fields": [
    { "name": "BrokerId", "type": "int32", "versions": "0+",
      "about": "The broker ID." },
	{ "name": "ClusterId", "type": "string", "versions": "0+",
	  "about": "The cluster id of the broker process." },
	{ "name": "IncarnationId", "type": "uuid", "versions": "0+",
	  "about": "The incarnation id of the broker process." },
    { "name": "CurrentMetadataOffset", "type": "int64", "versions": "0+",
      "about": "The highest metadata offset which the broker has reached." },
    { "name": "Listeners", "type": "[]Listener",
      "about": "The listeners of this broker", "versions": "0+", "fields": [
        { "name": "Name", "type": "string", "versions": "0+", "mapKey": true,
          "about": "The name of the endpoint." },
        { "name": "Host", "type": "string", "versions": "0+",
          "about": "The hostname." },
        { "name": "Port", "type": "uint16", "versions": "0+",
          "about": "The port." },
        { "name": "SecurityProtocol", "type": "int16", "versions": "0+",
          "about": "The security protocol." }
      ]
    { "name": "Features", "type": "[]Feature",
      "about": "The features on this broker", "versions": "0+", "fields": [
        { "name": "Name", "type": "string", "versions": "0+", "mapKey": true,
          "about": "The feature name." }
        { "name": "MinSupportedVersion", "type": "int16", "versions": "0+",
          "about": "The minimum supported feature level." },
        { "name": "MaxSupportedVersion", "type": "int16", "versions": "0+",
          "about": "The maximum supported feature level." }
      ]
    },
    { "name": "Rack", "type": "string", "versions": "0+", "nullableVersions": "0+",
      "about": "The rack which this broker is in." }
  ]
}

{
  "apiKey": 57,
  "type": "response",
  "name": "BrokerRegistrationResponse",
  "validVersions": "0",
  "flexibleVersions": "0+",
  "fields": [
    { "name": "ThrottleTimeMs", "type": "int32", "versions": "0+",
      "about": "Duration in milliseconds for which the request was throttled due to a quota violation, or zero if the request did not violate any quota." },
    { "name": "ErrorCode", "type": "int16", "versions": "0+",
      "about": "The error code, or 0 if there was no error." },
    { "name": "BrokerEpoch", "type": "int64", "versions": "0+", "default": "-1",
      "about": "The broker's assigned epoch, or -1 if none was assigned." }
  ]
}

BrokerHeartbeat

Required ACLs: CLUSTERACTION on CLUSTER

As described earlier, the broker periodically sends out a heartbeat request to the active controller.

Deprecated Configurations

...

New Error Codes

DUPLICATE_BROKER_REGISTRATION

There will be a new error code, DUPLICATE_BROKER_REGISTRATION, that the active controller will return when a broker tries to register with an ID that is currently in use.

INVALID_CLUSTER_ID

There will be a new error code, INVALID_CLUSTER_ID, that the controller will return if the broker tries to register with the wrong cluster ID.

RPCs

Obsoleting the Metadata Propagation RPCs

As discussed earlier, the new controller will use FetchRequest to fetch metadata from the active controller.  The details of how Raft fetching will work are spelled out in KIP-595: A Raft Protocol for the Metadata Quorum.

Since we propagate the metadata via Raft, we will no longer need to send out LeaderAndIsrRequest, UpdateMetadataRequest, and StopReplicaRequest.  These requests will be sent out only when we're in legacy mode, not when we're in KIP-500 mode.  Eventually we will add some support for these requests to the new controller, in order to support rolling upgrade from a pre-KIP-500 release. However, for the purpose of this KIP, the new controller will not use these requests.

Obsoleting the Controlled Shutdown RPC

The broker heartbeat mechanism replaces the controlled shutdown RPC.  Therefore, we will not need to support the this RPC any more in the controller-- except for compatibility during upgrades, which will be described further in a follow-on KIP.

...

{
  "apiKey": 5758,
  "type": "request",
  "name": "BrokerRegistrationRequestBrokerHeartbeatRequest",
  "validVersions": "0",
  "flexibleVersions": "0+",
  "fields": [
    { "name": "BrokerId", "type": "int32", "versions": "0+",
      "about": "The broker ID." },
    { "name": "BrokerEpoch", "type": "int64", "versions": "0+", "default": "0+-1",
      "about": "The broker IDepoch." },
	    { "name": "ClusterIdCurrentMetadataOffset", "type": "uuidint64", "versions": "0+",
	      "about": "The cluster id ofOne more than the highest metadata offset which the broker has processreached." },
	    { "name": "IncarnationIdWantFence", "type": "uuidbool", "versions": "0+",
	      "about": "The incarnation id of the broker processTrue if the broker wants to be fenced, false otherwise." },
    { "name": "NextMetadataOffsetWantShutDown", "type": "int64bool", "versions": "0+",
      "about": "TheTrue highestif metadatathe offsetbroker whichwants theto brokerinitiate hascontrolled not reachedshutdown." },
  ]
}   

{
  "nameapiKey": "Listeners"58,
  "type": "[]Listenerresponse",
    "name": "BrokerHeartbeatResponse",
  "aboutvalidVersions": "The listeners of this broker0",
  "versionsflexibleVersions": "0+",
  "fields": [
        { "name": "NameThrottleTimeMs", "type": "stringint32", "versions": "0+", "mapKey": true,
          "about": "TheDuration namein ofmilliseconds thefor endpoint." },
        { "name": "Host", "type": "string", "versions": "0+",
          "about": "The hostnamewhich the request was throttled due to a quota violation, or zero if the request did not violate any quota." },
        { "name": "PortErrorCode", "type": "int16", "versions": "0+",
      "about": "The error code, or 0 if there "about": "The portwas no error." },
        { "name": "SecurityProtocolIsCaughtUp", "type": "int16bool", "versions": "0+",
          "about": "The security protocol." }
      ]True if the broker has approximately caught up with the latest metadata." },
    { "name": "FeaturesIsFenced", "type": "bool", "versions": "[]Feature0+",
      "about": "TheTrue featuresif onthe thisbroker broker", "versions": "0+", "fields": [
    is fenced." },
    { "name": "NameShouldShutDown", "type": "stringbool", "versions": "0+",
      "mapKeyabout": true,
          "about": "The feature name." }
        { "name": "MinSupportedVersion", "type": "int16", "versions": "0+",
          "about": "The minimum supported feature level." },
        { "name": "MaxSupportedVersion", "type": "int16", "versions "True if the broker should proceed with its shutdown." }
  ]
}

The controller will wait to unfence a broker until it sends a heartbeat where ShouldFence is false and CurrentMetadataOffset is caught up.

If the heartbeat request has ShouldShutDown set, the controller will try to move all the leaders off of the broker.

The controller will set ControlledShutdownOk if the broker is cleared to execute a controlled shutdown.  In other words, if it has no leaderships.

The controller will return NOT_CONTROLLER if it is not active.  Brokers will always return NOT_CONTROLLER for these RPCs.

UnregisterBroker

Required ACLs: ALTER on CLUSTER

The UnregisterBrokerRequest asks the controller to unregister a broker from the cluster.

{
  "apiKey": 59,
  "type": "request",
  "name": "UnregisterBrokerRequest",
  "validVersions": "0",
  "flexibleVersions": "0+",
          "aboutfields": "The maximum supported feature level." }
      ]
    }, [
    { "name": "RackBrokerId", "type": "stringint32", "versions": "0+", "nullableVersions": "0+",
      "about": "The rackbroker whichID this broker is into unregister." }
  ]
}   

{
  "apiKey": 5759,
  "type": "response",
  "name": "BrokerRegistrationResponseUnregisterBrokerResponse",
  "validVersions": "0",
  "flexibleVersions": "0+",
  "fields": [
    { "name": "ThrottleTimeMs", "type": "int32", "versions": "0+",
      "about": "Duration in milliseconds for which the request was throttled due to a quota violation, or zero if the request did not violate any quota." },
    { "name": "ErrorCode", "type": "int16", "versions": "0+",
      "about": "The error code, or 0 if there was no error." },
    { "name": "BrokerEpoch", "type": "int64", "versions": "0+", "default": "-1",
      "about": "The broker's assigned epoch, or -1or 0 if nonethere was assignedno error." },
  ]
}

BrokerHeartbeat

As described earlier, the broker periodically sends out a heartbeat request to the active controller.

The valid response codes are:

• NONE if the unregistration succeeded or if the broker was already unregistered.
• NOT_CONTROLLER if the node that the request was sent to is not the controller
• UNSUPPORTED_VERSION if KIP-500 mode is not enabled

Record Formats

RegisterBrokerRecord

{
  "apiKey": 580,
  "type": "requestmetadata",
  "name": "BrokerHeartbeatRequestRegisterBrokerRecord"
,
  "validVersions": "0",
  "flexibleVersions": "0+",
  "fields": [
    	{ "name": "BrokerId", "type": "int32", "versions": "0+",
	      "about": "The broker IDid." },
    	{ "name": "BrokerEpochIncarnationId", "type": "int64uuid", "versions": "0+",
	  "defaultabout": "-1",
The incarnation     "about": "Theid of the broker epochprocess." },
    	{ "name": "CurrentMetadataOffsetBrokerEpoch", "type": "int64", "versions": "0+",
	      "about": "OneThe morebroker thanepoch theassigned highest metadata offset which the broker has reachedby the controller." },
    	{ "name": "ShouldFenceEndPoints", "type": "bool[]BrokerEndpoint", "versions": "0+",
   "nullableVersions": "0+",
	   "about": "TrueThe endpoints ifthat thecan brokerbe wantsused to becommunicate fenced,with falsethis otherwisebroker.", "fields": }
  ]
}   

{
  "apiKey": 58,
  "type": "response",
 [
		{ "name": "Name", "type": "string", "versions": "0+", "mapKey": true,
		  "about": "The name of the endpoint." },
		{ "name": "BrokerHeartbeatResponseHost",
  "validVersionstype": "0string",
  "flexibleVersionsversions": "0+",
		  "fieldsabout": [
    "The hostname." },
		{ "name": "ThrottleTimeMsPort", "type": "int32uint16", "versions": "0+",
  		    "about": "DurationThe in milliseconds for which the request was throttled due to a quota violation, or zero if the request did not violate any quota." },
    port." },
		{ "name": "SecurityProtocol", "type": "int16", "versions": "0+",
		  "about": "The security protocol." }
	]},
	{ "name": "ErrorCodeFeatures", "type": "int16[]BrokerFeature", "versions": "0+",
    "nullableVersions": "0+",
	  "about": "The errorfeatures code,that orthis 0 if there was no errorbroker supports." },
    "fields": [
		{ "name": "IsCaughtUpName", "type": "boolstring", "versions": "0+",
    "mapKey": true,
		  "about": "TrueThe if the broker has approximately caught up withname of the latest metadatafeature." },
    		{ "name": "IsFencedMinVersion", "type": "boolint16", "versions": "0+",
   		   "about": "True if theThe minimum feature level that this broker is fencedsupports." },
  ]
}

The controller will wait to unfence a broker until it has sent at least one heartbeat where that broker's currentState is active.  So a typical transition will look like this:

• broker sends a BrokerHeartbeatRequest with currentState = fenced, targetState = active.  controller response with nextState = active
• broker sends a BrokerHeartbeatRequest with currentState = active, targetState = active.  controller response with nextState = active
• controller unfences the broker

The second step informs the controller that the broker has received its response to the first step.  In order to speed this process along, the broker will send the second heartbeat immediately after receiving the first, rather than waiting for the heartbeat timeout.

As always with enums, the UNKNOWN state is used only to translate values that our software is too old to understand.

The controller will return NOT_CONTROLLER if it is not active.  Brokers will always return NOT_CONTROLLER for these RPCs.

Heartbeat requests that give a NextMetadataOffset which is lower than the previous one recorded for that broker will be ignored.

Record Formats

RegisterBrokerRecord

{
  "apiKey": 0,
 		{ "name": "MaxVersion", "type": "int16", "versions": "0+",
		  "about": "The maximum feature level that this broker supports." }
	]},
	{ "name": "Rack", "type": "metadatastring",
  "nameversions": "RegisterBrokerRecord0+"
,
  "validVersionsnullableVersions": "0+",
	  "fieldsabout": [
	 "The broker rack." }
  ]
}

UnregisterBrokerRecord

{
  "apiKey": 1,
 { "name": "Id", "type": "int32metadata",
  "versionsname": "0+UnregisterBrokerRecord",
	  "aboutvalidVersions": "0"The,
 broker id."fields": },[
	{ "name": "IncarnationIdBrokerId", "type": "uuidint32", "versions": "0+",
	  "about": "The incarnationbroker id of the broker process." },
	{ "name": "BrokerEpoch", "type": "int64", "versions": "0+",
	  "about": "The broker epoch." assigned}
 by the controller." },
	{ "name": ]
}

TopicRecord

{
  "apiKey": 2,
  "EndPoints", "type": "[]BrokerEndpointmetadata",
  "versionsname": "0+TopicRecord",
  "nullableVersionsvalidVersions": "0+",
	  "aboutfields": "The endpoints that can be used to communicate with this broker.", "fields": [
		[
        { "name": "NameTopicName", "type": "string", "versions": "0+", "mapKey": true,
		
          "about": "The name of the endpointtopic name." },
		        { "name": "HostTopicId", "type": "stringuuid", "versions": "0+",
		          "about": "The hostname unique ID of this topic." }
  ]
}

PartitionRecord

{
  "apiKey,
		{ "name": "Port"3,
  "type": "int16metadata",
  "versionsname": "0+PartitionRecord",
		  "aboutvalidVersions": "0"The port." },
		,
  "fields": [
    { "name": "SecurityProtocolPartitionId", "type": "int16int32", "versions": "0+",
		 "default": "-1",
      "about": "The securitypartition protocolid." }
	]},
	    { "name": "FeaturesTopicId", "type": "[]BrokerFeatureuuid", "versions": "0+", "nullableVersions": "0+",
	
      "about": "The unique featuresID thatof this broker supportstopic." }, "fields": [
		
    { "name": "NameReplicas", "type":  "string[]int32", "versions":  "0+", "mapKey": true,
		
      "about": "The namereplicas of the feature this partition, sorted by preferred order." },
		    { "name": "MinVersionIsr", "type":  "int16[]int32", "versions":  "0+",
		      "about": "The minimumin-sync featurereplicas level thatof this broker supports.partition" },
		    { "name": "MaxVersionRemovingReplicas", "type":  "int16[]int32", "versions":  "0+",
		      "about": "The replicas that maximumwe featureare levelin thatthe thisprocess brokerof supportsremoving." }
	]},
	    { "name": "RackAddingReplicas", "type":  "string[]int32", "versions":  "0+", "nullableVersions": "0+",
	
      "about": "The broker rack replicas that we are in the process of adding." },
    ]
}

UnregisterBrokerRecord

{
  "apiKeyname": 1"Leader",
  "type": "metadataint32",
  "nameversions": "UnregisterBrokerRecord0+",
  "validVersionsdefault": "0-1",
   "fields": [
	{   "nameabout": "Id", "type": "int32", "versions": "0+",
	  "about": "The broker idThe lead replica, or -1 if there is no leader." },
	    { "name": "EpochLeaderEpoch", "type": "int64int32", "versions": "0+",
	 "default": "-1",
      "about": "The broker epochAn epoch that gets incremented each time we change the leader." }
  ]
}

...

ConfigRecord

{
  "apiKey": 2,
  "type": "metadata",
  "name": "TopicRecord",
  "validVersionsapiKey": "0"4,
  "fieldstype": [
        { "name": "Name", "type": "string", "versions": "0+",
          "about": "The topic name." },
     "metadata",
  "name": "ConfigRecord",
  "validVersions": "0",
  "fields": [
    { "name": "TopicIdResourceType", "type": "uuidint8", "versions": "0+",
          "about": "The uniquetype IDof ofresource this configuration applies topicto." },
        { "name": "DeletingResourceName", "type": "booleanstring", "versions": "0+",
          "about": "TrueThe ifname thisof topicthe isresource inthis theconfiguration process of being deletedapplies to." }
  ]
}

PartitionRecord

{
  "apiKey": 3,
  "type": "metadata",
  "name": "PartitionRecord",
  "validVersions": "0",
    "fields": [
    { "name": "PartitionIdName", "type": "int32string", "versions": "0+", "default": "-1",
      "about": "The partition id name of the configuration key." },                  
    { "name": "TopicIdValue", "type": "uuidstring", "versions": "0+",     
      "about": "The unique IDvalue of thisthe topicconfiguration." },
  ]           
} 

PartitionChangeRecord

{
  "nameapiKey": "Replicas"5,
  "type":  "[]int32metadata",
  "versionsname":  "0+",
      "about": "The replicas of this partition, sorted by preferred order." },PartitionChangeRecord",
  "validVersions": "0",
  "fields": [
    { "name": "IsrPartitionId", "type":  "[]int32", "versions":  "0+", "default": "-1",
      "about": "The in-sync replicas of this partitionpartition id." },
    { "name": "RemovingReplicasTopicId", "type":  "[]int32uuid", "versions":  "0+",
      "about": "The replicasunique thatID weof are in the process of removingthis topic." },
    { "name": "AddingReplicasIsr", "type":  "[]int32", "versionsdefault":  "0+null",
      "aboutentityType": "The replicas that we are in the process of adding." },
brokerId",
     { "nameversions": "Leader0+", "typenullableVersions": "int320+", "versionstaggedVersions": "0+", "defaulttag": "-1"0,
      "about": "Thenull if leadthe replica, or -1 if there is no leaderISR didn't change; the new in-sync replicas otherwise." },
    { "name": "LeaderEpochLeader", "type": "int32", "default": "-2", "entityType": "brokerId",
      "versions": "0+", "defaulttaggedVersions": "-1"0+", "tag": 1,
      "about": "An-1 epochif thatthere getsis incrementednow eachno timeleader; we-2 changeif the leader." }
  ]
}

ConfigRecord

{      didn't change; the new leader otherwise." },
    
{  "apiKeyname": 4"Replicas",
  "type": "metadata[]int32",
  "namedefault": "ConfigRecordnull",
  "validVersionsentityType": "0brokerId",
  "fields": [
    { "nameversions": "ResourceType0+", "typenullableVersions": "int80+", "versionstaggedVersions": "0+", "tag": 2,
      "about": "The type of resource this configuration applies tonull if the replicas didn't change; the new replicas otherwise." },
    { "name": "ResourceNameRemovingReplicas", "type": "string[]int32", "versionsdefault": "0+null",
      "aboutentityType": "The name of the resource this configuration applies to." },         
brokerId",
     { "nameversions": "Name0+", "typenullableVersions": "string0+", "versionstaggedVersions": "0+", "tag": 3,
      "about": "Thenull nameif ofthe theremoving configuration key." },      replicas didn't change; the new removing replicas otherwise." },
    { "name": "AddingReplicas", "type": "[]int32", "default": "null", "entityType": "brokerId",
     { "nameversions": "Value0+", "typenullableVersions": "string0+", "versionstaggedVersions": "0+",     "tag": 4,
      "about": "Thenull value ofif the configuration." }
  ]           adding replicas didn't change; the new adding replicas otherwise." }
  ]
} 

...

AccessControlRecord

{
  "apiKey": 56,
  "type": "metadata",
  "name": "IsrChangeRecordAccessControlRecord",
  "validVersions": "0",
  "fields": [
    { "name": "PartitionIdResourceType", "type": "int32int8", "versions": "0+", "default": "-1",
      "about": "The partitionresource id.type" },
    { "name": "TopicIdResourceName", "type": "uuidstring", "versions": "0+", "nullableVersions": "0+",
      "about": "The resource name, uniqueor IDnull ofif this topic is for the default resource." },
    { "name": "IsrPatternType", "type":  "[]int32int8", "versions":  "0+",
      "about": "The in-syncpattern replicastype of this partition(literal, prefixed, etc.)" },
    { "name": "LeaderPrincipal", "type": "int32string", "versions": "0+", "default": "-1",
      "about": "The lead replica, or -1 if there is no leaderprincipal name." },
    { "name": "LeaderEpochHost", "type": "int32string", "versions": "0+", "default": "-1",
      "about": "An epoch that gets incremented each time we change the leaderThe host." },
   ]
} 

AccessControlRecord

{
  "apiKeyname": 6"Operation",
  "type": "metadataint8",
  "nameversions": "AccessControlRecord0+",
      "validVersionsabout": "0",
The operation type."fields": [},
    { "name": "ResourceTypePermissionType", "type": "int8", "versions": "0+",
      "about": "The resourcepermission type (allow, deny)." },
  ]
} 

FenceBrokerRecord

{
 { "nameapiKey": "ResourceName"7,
  "type": "string", "versions": "0+metadata",
  "nullableVersionsname": "0+FenceBrokerRecord",
      "aboutvalidVersions": "The resource name, or null if this is for the default resource." },0",
  "fields": [
    { "name": "PatternTypeBrokerId", "type": "int8int32", "versions": "0+",
      "about": "The pattern type (literal, prefixed, etc.) broker ID to fence. It will be removed from all ISRs." },
    { "name": "PrincipalBrokerEpoch", "type": "stringint64", "versions": "0+",
      "about": "The principal name epoch of the broker to fence." },
  ]
} 

UnfenceBrokerRecord

{
 { "nameapiKey": "Host"8,
  "type": "stringmetadata",
  "versionsname": "0+UnfenceBrokerRecord",
      "aboutvalidVersions": "0"The host." },,
  "fields": [
    { "name": "OperationBrokerId", "type": "int8int32", "versions": "0+",
      "about": "The broker ID operationto typeunfence." },
    { "name": "PermissionTypeBrokerEpoch", "type": "int8int64", "versions": "0+",
      "about": "The epoch permissionof the typebroker (allow, deny)to unfence." }
  ]
} 

...

RemoveTopic

{
  "apiKey": 79,
  "type": "metadata",
  "name": "FenceBrokerRecordRemoveTopicRecord",
  "validVersions": "0",
  "fields": [
    { "name": "Id", "type": "int32", "versions": "0+",
      "about": "The broker ID to fence. It will be removed from all ISRs."fields": }[
    { "name": "EpochTopicId", "type": "int64uuid", "versions": "0+",
      "about": "The epoch of the broker to fence topic to remove. All associated partitions will be removed as well." }
  ]
} 

...

DelegationTokenRecord

{
  "apiKey": 810,
  "type": "metadata",
  "name": "UnfenceBrokerRecordDelegationTokenRecord",
  "validVersions": "0",
  "fields": [
    { "name": "IdOwner", "type": "int32string", "versions": "0+",
      "about": "The brokerdelegation IDtoken to unfenceowner." },
    { "name": "EpochRenewers", "type": "int64[]string", "versions": "0+",
      "about": "The epochprincipals of the broker to unfence." }
  ]
} 

RemoveTopic

{
  "apiKey": 9,
  "type": "metadata",
  "name": "RemoveTopicRecord",
  "validVersions": "0",
  "fields": [which have renewed this token." },
    { "name": "IdIssueTimestamp", "type": "uuidint64", "versions": "0+",
      "about": "The topictime toat remove.which Allthis associatedtimestamp partitions will be removed as wellwas issued." },
   ]
} 

DelegationTokenRecord

{
  "apiKeyname": 10"MaxTimestamp",
  "type": "metadataint64",
  "nameversions": "DelegationTokenRecord0+",
      "validVersionsabout": "0",
  "fields": [The time at which this token cannot be renewed any more." },
    { "name": "OwnerExpirationTimestamp", "type": "stringint64", "versions": "0+",
      "about": "The delegation next time at which this token must be ownerrenewed." },
    { "name": "RenewersTokenId", "type": "[]string", "versions": "0+",
      "about": "The principals which have renewed this tokentoken id." },
  ]
} 

UserScramCredentialRecord

{
 { "nameapiKey": "IssueTimestamp"11,
  "type": "int64metadata",
  "versionsname": "0+UserScramCredentialRecord",
      "aboutvalidVersions": "The time at which this timestamp was issued." },"0",
  "fields": [
    { "name": "MaxTimestampUserName", "type": "int64string", "versions": "0+",
      "about": "The time at which this token cannot be renewed any moreuser name." },
    { "name": "ExpirationTimestampCredentialInfos", "type": "int64[]CredentialInfo", "versions": "0+",
      "about": "The mechanism nextand timerelated atinformation whichassociated thiswith tokenthe mustuser's beSCRAM renewedcredential.", "fields": },[
      { "name": "TokenIdMechanism", "type": "stringint8", "versions": "0+",
        "about": "The tokenSCRAM idmechanism." },
    ]
} 

UserScramCredentialRecord

{
   { "apiKeyname": 11"Salt",
  "type": "metadatabytes",
  "nameversions": "UserScramCredentialRecord0+",
        "validVersionsabout": "0"A random salt generated by the client." },
  "fields": [
    { "name": "NameSaltedPassword", "type": "stringbytes", "versions": "0+",
        "about": "The usersalted namepassword." },
      { "name": "CredentialInfosIterations", "type": "[]CredentialInfoint32", "versions": "0+",
        "about": "The mechanism and related information associated with the user's SCRAM credential.", "fields": [
      { "name" number of iterations used in the SCRAM credential." }]}
  ]
} 

FeatureLevelRecord

{
  "apiKey": 12,
  "type": "metadata",
  "name": "FeatureLevelRecord",
  "validVersions: "Mechanism", "type": "int8", "versions": "0+",
        "aboutfields": "The SCRAM mechanism." },
  [
    { "name": "SaltName", "type": "bytesstring", "versions": "0+",
 "mapKey": true,
      "about": "AThe random salt generated by the clientfeature name." },
      { "name": "SaltedPasswordMinFeatureLevel", "type": "bytesint16", "versions": "0+",
        "about": "The salted password current finalized minimum feature level of this feature for the cluster." },
      { "name": "IterationsMaxFeatureLevel", "type": "int32int16", "versions": "0+",
        "about": "The number": "The current finalized maximum feature level of iterationsthis usedfeature infor the SCRAM credentialcluster." }]}
  ]
} 

...

FailedReplicasRecord

{
  "apiKey": 1213,
  "type": "metadata",
  "name": "FeatureLevelRecordFailedReplicasRecord",
  "validVersions": "0",
  "fields": [
    { "name": "NameBrokerId", "type": "stringint32", "versions": "0+", "mapKey": true,
      "about": "The featurebroker nameid." },
    { "name": "MinFeatureLevelTopics", "type": "int16[]TopicWithFailures", "versions": "0+",
      "about": "The current finalized minimum feature level of this feature for the cluster topics with failed replicas.", "fields": [
      { "name": "TopicId", "type": "uuid", "versions": "0+",
        "about": "The topic UUID." },
      { "name": "MaxFeatureLevelPartitions", "type": "int16[]int32", "versions": "0+",
        "about": "The currentpartition finalizedids." maximum} feature level of this
 feature for the cluster." ]}
  ]
} 

...

QuotaRecord

{
  "apiKey": 1314,
  "type": "metadata",
  "name": "FailedReplicasRecordQuotaRecord",
  "validVersions": "0",
  "fields": [
    { "name": "Entity", "type": "[]EntityData", "versions": "0+",
      "about": "The quota entity to alter.", "fields": [
      { "name": "BrokerEntityType", "type": "int32string", "versions": "0+",
        "about": "The brokerentity idtype." },
      { "name": "TopicEntityName", "type": "uuidstring", "versions": "0+", "nullableVersions": "0+",
        "about": "The topic UUID." }name of the entity, or null if the default." }
    ]},
    { "name": "PartitionsKey", "type": "[]int32string", "versions": "0+",
      "about": "The quota partitionconfiguration idskey." },
    ]
} 

New Metrics

...

kafka.controller:type=KafkaController,name=MetadataLag

...

The offset delta between the latest metadata record this controller has replayed and the last stable offset of the metadata topic.

...

New name for kafka.controller:type=KafkaController,name=SnapshotLag

The offset delta between the latest stable offset of the metadata topic and the offset of the last snapshot (or the last stable offset itself, if there are no snapshots)

{ "name": "Value", "type": "float64", "versions": "0+",
      "about": "The value to set, otherwise ignored if the value is to be removed." },
    { "name": "Remove", "type": "bool", "versions": "0+",
      "about": "Whether the quota configuration value should be removed, otherwise set." }
  ]
} 

New Metrics

Unused Metrics in KIP-500 Mode

...

Full Name	Description
kafka.servercontroller:type=SessionExpireListenerKafkaController,name=ZooKeeperExpiresPerSec	No longer needed when running in KIP-500 mode because we won't have any ZK sessions

Compatibility, Deprecation, and Migration Plan

As described above, this KIP outlines a new mode that the broker can run in, KIP-500 mode.  For now, this mode will be experimental, and there will be no way to migrate existing clusters from legacy mode to KIP-500 mode.  We plan on outlining how this upgrade process will work in a follow-on KIP.  We do plan on deprecating legacy mode eventually, but we are not quite ready to do it yet in this KIP.

Since KIP-500 mode is currently in a pre-alpha state, we do not guarantee that future versions will support upgrading from the current version of it yet.  Once it is more stable, we will have a more traditional binary compatibility regime.

Rejected Alternatives

Suport Automatic Broker ID Assignment

This KIP proposes to drop support for automatic broker ID assignment.  What if we decided to continue to support it?

If we were willing to take a little bit more complexity on board, it would be relatively easy to support automatic broker ID assignment.  Brokers could simply ask the active controller to assign them a new ID when starting up, just as they previously obtained one from ZooKeeper.

However, automatic controller ID assignment is a much more difficult problem.  We never actually supported automatically assigning ZooKeeper IDs, so there is no pattern to follow here.  In general, Raft assumes that nodes know their IDs before the protocol begins. We cannot rely on random assignment because the 31 bit space is not large enough. We could perhaps create a separate protocol for assigning node IDs, but it might be complex.  

In general it's not clear how useful automatic broker ID assignment really is.  Configuration management software like Puppet, Chef, or Ansible can easily create a new ID for each node's configuration file.  Therefore, it's probably best to use this compatibility break to drop support for automatic broker ID assignment.

Combined Heartbeats and Fetch Requests

The brokers are always fetching new metadata from the controller.  Why not combine these fetch requests with the heartbeat requests, so that the brokers only have to send one request rather than two?

The main reason for making them separate requests is to have better separation of concerns.  Fetching metadata is logically a bit different than sending a heartbeat, and coupling them could result in a messy design and code.  We would have to add significant extra complexity to the FetchRequest schema.  Perhaps even worse, we would need to make the timing of fetch requests line up with the timing needed for broker heartbeats.

Shared IDs between Multiple Nodes

One possibility that we've discussed is whether we could have controller IDs and broker IDs share the same ID space.  For example, could there be both a broker 1 and a controller 1?  This is not a good idea because NetworkClient assumes a single ID space.  So if there is both a controller 1 and a broker 1, we don't have a way of picking the "right" one.  We would have to add a concept of node types.  That is a fair amount of extra work.  It also is a bit conceptually messy for the network layer to understand node types, rather than just treating them all as endpoints.

Unused Metrics in KIP-500 Mode

We will deprecate these metrics as soon as legacy mode is deprecated.  For now, they will be unused in KIP-500 mode.

Compatibility, Deprecation, and Migration Plan

As described above, this KIP outlines a new mode that the broker can run in, KIP-500 mode.  For now, this mode will be experimental, and there will be no way to migrate existing clusters from legacy mode to KIP-500 mode.  We plan on outlining how this upgrade process will work in a follow-on KIP.  We do plan on deprecating legacy mode eventually, but we are not quite ready to do it yet in this KIP.

Since KIP-500 mode is currently in a pre-alpha state, we do not guarantee that future versions will support upgrading from the current version of it yet.  Once it is more stable, we will have a more traditional binary compatibility regime.

Rejected Alternatives

Suport Automatic Broker ID Assignment

This KIP proposes to drop support for automatic broker ID assignment.  What if we decided to continue to support it?

If we were willing to take a little bit more complexity on board, it would be relatively easy to support automatic broker ID assignment.  Brokers could simply ask the active controller to assign them a new ID when starting up, just as they previously obtained one from ZooKeeper.

However, automatic controller ID assignment is a much more difficult problem.  We never actually supported automatically assigning ZooKeeper IDs, so there is no pattern to follow here.  In general, Raft assumes that nodes know their IDs before the protocol begins. We cannot rely on random assignment because the 31 bit space is not large enough. We could perhaps create a separate protocol for assigning node IDs, but it might be complex.  

In general it's not clear how useful automatic broker ID assignment really is.  Configuration management software like Puppet, Chef, or Ansible can easily create a new ID for each node's configuration file.  Therefore, it's probably best to use this compatibility break to drop support for automatic broker ID assignment.

Combined Heartbeats and Fetch Requests

The brokers are always fetching new metadata from the controller.  Why not combine these fetch requests with the heartbeat requests, so that the brokers only have to send one request rather than two?

The main reason for making them separate requests is to have better separation of concerns.  Fetching metadata is logically a bit different than sending a heartbeat, and coupling them could result in a messy design and code.  We would have to add significant extra complexity to the FetchRequest schema.  Perhaps even worse, we would need to make the timing of fetch requests line up with the timing needed for broker heartbeatsA related question is whether a broker and a controller could share the same port, if they are co-located in the same JVM.  We discussed this as part of KIP-590 and agreed that we would not share ports.  A shared port would make it impossible to have separate RPC handlers, complicating the code.  It would also make it impossible to have separate, stricter authentication for controllers.  Sharing a port with the broker opens us up to denial of service attacks, including unintentional ones.