This page displays the current state of the cluster.

Cluster/Node status

Nodes have the following possible statuses:

Status FamilyStatusDescription Status in the JSON nodes API
Initializing processWaiting for initial configurationnode waits until the primary-node address is set
Startingnode handles bootstrap activities
Waiting for initializationnode waits for the cluster manager (the primary master) to order to transition to INITIALIZE state
Initializingnode starts services required to be able to connect to other nodes
Connecting to other nodesnode connects to all other nodes
Loading warehousenode loads files from the warehouse (metadata on schemas and tables and indexes)
Ready to migrate warehousenode waits for the cluster manager to order to transition to MIGRATE state
Migrating warehousenode migrates the warehouse if a migration process is required after an upgrade
Ready to servenode waits for the cluster manager to order the transition to SERVE state
ServingServingnode is open and allowed to answer incoming queries
On errorOn errornode is in error and won't accept incoming queries

An initializing cluster will display the status of each connected node:

A cluster in error will provide an error message describing the problem:

Indexima cluster is available for SQL queries when all nodes are in status "Serving":

Note that only at startup, only the recently used indexes are loaded into memory. Unused indexes won't be loaded to nodes until they are first used. See parameter table.prefetch.lastdays in galactica.conf.

Re-initialize cluster

  • The button RE-INITIALIZE CLUSTER interrupts nodes communication:
    • All running queries are killed
    • The nodes move to the initializing status
    • The warehouse is not reloaded if the cluster topology hasn't changed
  • Select the option "Reload Warehouse" to force the cluster to perform a full reload of the tables and indexes from the warehouse.