Thank you for your answer Jörg. I remove them in conf file but my performance problem continues. I have another question about it. My network stat is that;
curl -XGET 'http://localhost:9200/_nodes/stats/network?human&pretty' { "cluster_name" : "test-cluster", "nodes" : { "XB95yJZhS7WLAPBm9O994Q" : { "timestamp" : 1411643924755, "name" : "test", "transport_address" : "inet[/127.0.0.1:9300]", "host" : "test-host", "ip" : [ "inet[/127.0.0.1:9300]", "NONE" ], "attributes" : { "master" : "true" }, "network" : { "tcp" : { "active_opens" : 794549, "passive_opens" : 489890, "curr_estab" : 402, "in_segs" : 99304476, "out_segs" : 100905758, "retrans_segs" : 53037, "estab_resets" : 10989, "attempt_fails" : 420884, "in_errs" : 18226, "out_rsts" : 439001 } } } } } Have you any idea about why curr_estab value is so big? I use elasticsearch.py for insertion and only 8 connections are active for indexing data on 2 processes. On Thursday, September 25, 2014 11:21:35 AM UTC+3, Jörg Prante wrote: > > Do not manipulate the threadpool to "fixed" and do not use such high > numbers like 100,600, 5000, 50000.... This will sooner or later congest > your machine. The long list of threads in OS is just one (harmless) symptom > of a misconfiguration. Use the default setting. > > Do not use 80% buffer for index. This hurts search performance and cache > resources. Use the default setting. > > Do not use cache field type "soft". This hides cache and GC problems and > gives cryptic exceptions plus very bad performance in spikes. Use the > default setting. > > I hope the single node is a development machine. For better performance, > use at least 3 nodes on 3 machines. > > Jörg > > >> indices.memory.index_buffer_size: 80% >> index.store.compress.stored: true >> index.store.fs.lock: none >> >> >> threadpool.search.type: fixed >> threadpool.search.size: 600 >> threadpool.search.queue_size: 10000 >> >> >> threadpool.bulk.type: fixed >> threadpool.bulk.size: 600 >> threadpool.bulk.queue_size: 10000 >> >> >> threadpool.index.type: fixed >> threadpool.index.size: 100 >> threadpool.index.queue_size: 5000 >> >> >> index.cache.field.type: soft >> index.cache.field.max_size: 50000 >> index.cache.field.expire: 24h >> >> > -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/82a3d79b-19a4-4493-aab2-0dcfdd062bbe%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.