Skip to content
QuerybaraQuerybara

Cluster

  • Elasticsearch

The cluster panel answers the questions that come up when a cluster turns yellow or red: which nodes are there, which shards are not started and why, how full each disk is, and what is running. Double-click Cluster in the sidebar to open it. Refresh in its toolbar reads everything again.

The Elasticsearch cluster's Shards tab with an unassigned replica of products-v2, and the allocation explanation: the only node refuses it because a copy of the shard is already there (same_shard).The Elasticsearch cluster's Shards tab with an unassigned replica of products-v2, and the allocation explanation: the only node refuses it because a copy of the shard is already there (same_shard).
Find unassigned shards and read why the cluster cannot place them.

Health and nodes

Section titled “Health and nodes”

The Health and nodes tab shows the cluster name and status, the server version, the number of nodes (and data nodes), active shards with the share that is active, and the relocating, initializing and unassigned shards and pending tasks.

The node table lists each node’s name, address, roles, heap, CPU, load, disk and version.

Shards

Section titled “Shards”

The Shards tab lists every shard copy with its index, shard number, copy, state, node, documents and size. Unassigned shards come first.

  • Filter by index narrows the list.
  • Only shards that are not started hides the healthy ones.
  • Explain on a row, or Explain the first unassigned, asks the cluster why a shard is where it is.

The explanation shows the shard’s state and node, why it is unassigned and the details, whether it can be allocated, and each node’s decision with the reason.

Disk

Section titled “Disk”

The Disk tab shows the disk watermarks as the cluster reads them, including a node’s own elasticsearch.yml:

Watermark Meaning shown
Low (no new replicas above) No new replicas go to a node above it
High (shards move away above) Shards move off a node above it
Flood stage (indices turn read-only above) Indices on the node turn read-only

Where the cluster sets a headroom cap, it is shown beside the percentage (“95%, or once less than 100GB is free”). Threshold checks says whether the watermarks are enforced at all. The node disk table shows each node’s shards, used share, used and total, and free space, with the watermarks marked on the bars. Watermarks given as free space rather than percentages are listed but not drawn on the bars, since they depend on each disk’s size.

Tasks

Section titled “Tasks”

The Tasks tab lists running tasks with their id, action, running time, progress and description. Cancel… cancels a cancellable task. Production and confirm-writes profiles ask first; a read-only profile cannot cancel.

Section titled “Related”

Documents Querybara 0.1.1 · built frombc9f5aa