rpk: command(s) for cancelling partition movement

[ztlpn]

EITHER rpk cluster partitions move --cancel OR rpk cluster partitions movement-cancel.

Scenario 1: cancel all partition movements in the cluster

Corresponding admin API endpoint: POST /v1/cluster/cancel_reconfigurations.

Query parameters: none

Response: a list of partition_result objects.

Response example:

[
  {
    "ns": "kafka",
    "topic": "foo",
    "partition": 1,
    "result": "cluster::errc:0"
  },
  {
    "ns": "kafka",
    "topic": "foo",
    "partition": 4,
    "result": "cluster::errc:0"
  }
]

In case of error a standard admin API response is returned, example:

{
  "message": "Unexpected error: rpc::errc::exponential_backoff",
  "code": 500
}

Scenario 2: cancel all partition movements on a node

Corresponding admin API endpoint: POST /v1/brokers/{id}/cancel_partition_moves.

Query parameters:

• id - node id.

Response: the same as in scenario 1.

Scenario 3 (low priority): cancel a single partition movement

Corresponding admin API endpoints

• to safely cancel: POST /v1/partitions/{namespace}/{topic}/{partition}/cancel_reconfiguration
• to forcibly abort: POST /v1/partitions/{namespace}/{topic}/{partition}/unclean_abort_reconfiguration

Query parameters:

• namespace
• topic
• partition

Response: empty if successful, otherwise standard admin API error response.

Note: force-abort versions of scenario 1 and scenario 2 need to be implemented too, but redpanda support is not ready yet (tracking in #5782)

[ztlpn]

cc @mmaslankaprv