Hi
I'm running several tests against a Elasticsearch 2.1. I created five indexes with one shard each one and start to insert in http//IP:9200/_bulk API point. I'm using a python client using the request module , inserting 100 docs each request. Well, system performs well (from the return JSON I see the took key about 90 100 ms). I hit the ctrl-c and start again the test. Drop the indexes and start to insert the same events ... well, after this, I see that the time to insert multiply by 10. With the same script and the same date.
I check the "hot threads" and see these stacktrace:
root@elasticnew:/etc/elasticsearch# curl -XGET '172.31.2.191:9200/_nodes/hot_threads?threads=3&type=wait'
::: {Luchino Nefaria}{uH0reNQjSpKuvg2ux0FZvA}{172.31.2.191}{172.31.2.191:9300}
Hot threads at 2015-12-13T22:41:16.744Z, interval=500ms, busiestThreads=3, ignoreIdleThreads=true:
0.0% (0s out of 500ms) wait usage by thread 'elasticsearch[keepAlive/2.1.0]'
10/10 snapshots sharing following 8 elements
sun.misc.Unsafe.park(Native Method)
java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:836)
java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireSharedInterruptibly(AbstractQueuedSynchronizer.java:997)
java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireSharedInterruptibly(AbstractQueuedSynchronizer.java:1304)
java.util.concurrent.CountDownLatch.await(CountDownLatch.java:231)
org.elasticsearch.bootstrap.Bootstrap$1.run(Bootstrap.java:68)
java.lang.Thread.run(Thread.java:745)
root@elasticnew:/etc/elasticsearch#
I don't know why the second time the system slowdown .... :?
The script I'm using is genloadindexes.py.
I launch it with
./genloadindexes.py http://IP:9200/ 1M.events 5
First time, each insert request is fine. But If I stop and start to insert the same data ... the system slows ...