Change the breaker limit¶
This is a temporary change, and probably a dumb one.
curl -XPUT localhost:9200/_cluster/settings -d '{
"transient" : {
"indices.breaker.request.limit" : "41%"
}
}'
get info about nodes in cluster¶
sort by role, then disk usage
GET _cat/nodes?v&h=name,role,master,shardStatsTotalCount,diskTotal,diskUsedPercent,cpu,load_1m,load_5m,load_15m,heapMax,heapCurrent,heapPercent&s=role,diskUsedPercent
get pipeline statistics for troubleshooting ingest processing times¶
GET _nodes/stats/ingest?filter_path=nodes.*.ingest
get number of indexing failures for all indices¶
GET _all/_stats/indexing?filter_path=indices.*.total.indexing.index_failed
show the current write thread pool statistics¶
queue is particularly useful for diagnosing the overall write load on the cluster see: cat-thread-pool
GET _cat/thread_pool/write?v&h=node_name,name,active,size_queue,queue_size,rejected