ALTER CLUSTER

Alter the state of an existing cluster.

Table of Contents

Synopsis

  1. ALTER CLUSTER
  2.   { REROUTE RETRY FAILED
  3.   | DECOMMISSION <nodeId | nodeName>
  4.   | SWAP TABLE source TO target [ WITH ( expr = expr [ , ... ] ) ]
  5.   | GC DANGLING ARTIFACTS
  6.   }

Description

ALTER CLUSTER applies a change to the cluster state.

Arguments

REROUTE RETRY FAILED

The index setting allocation.max_retries indicates the maximum of attempts to allocate a shard on a node. If this limit is reached it leaves the shard unallocated. This command allows the enforcement to retry the allocation of shards which failed to allocate. See Reroute Shards to get convenient use-cases.

The rowcount defines the number of shards that will be allocated. A rowcount of -1 reflects an error or indicates that the statement did not get acknowledged.

Note

This statement can only be invoked by superusers that already exist in the cluster.

Additionally, keep in mind that this statement only triggers the shard re-allocation and is therefore asynchronous. Unassigned shards with large size will take some time to allocate.

DECOMMISSION <nodeId | nodeName>

This command triggers a graceful cluster node decommission. The node can be specified by either its Id or name. See Graceful stop for more information on decommissioning nodes gracefully.

SWAP TABLE

  1. SWAP TABLE source TO target [ WITH ( expr = expr [ , ... ] ) ]

This command swaps two tables. source will be renamed to target and target will be renamed to source.

An example use case of this feature is some sort of schema migration using INSERT INTO ... query. You’d create a new table with an updated schema, copy over the data from the old table and then replace the old table with the new table.

Note

Swapping two tables causes the shards to be unavailable for a short period.

Options

drop_source

Default: false

A boolean option that if set to true causes the command to remove the source table after the rename. This causes the command to replace source with target, instead of swapping the names.

GC DANGLING ARTIFACTS

  1. GC DANGLING ARTIFACTS

Some operations in CrateDB might temporarily create data to complete the operation. If during such an operation the cluster starts failing these temporary artifacts might not be cleaned up correctly.

The ALTER CLUSTER GC DANGLING ARTIFACTS command can be used to remove all artifacts created by such operations.