Overview
The management queue is where tasks such as node allocation or index management tasks are queued if they cannot be carried out immediately. In a stable cluster, it would be normal to have one management thread per node, with no rejections. If management tasks start to back up, it’s an indication that:
- An excessive number of management tasks are being created, or
- Something is stopping the management tasks from being carried out properly.
A high number of tasks in the management queue can cause cluster instability which could result in shards not being allocated, possible loss of data and your cluster turning red. The cause should be investigated and resolved as soon as possible.
How to resolve the issue
To see the status of the thread queue, you can run the following command:
GET /_cat/thread_pool/management?v&h=id,active,rejected,completed,node_id
This should give you an idea of the current status of the thread pool, and which node(s) have tasks being queued.
You can also run this command:
GET /_cat/pending_tasks?v
This should give you an indication of what sort of tasks are being queued on the cluster.
