|
In JForex, there is a great tool for optimization strategy parameters. I am trying to develop my own system, very similar to that of JForex but unlike JForex, my system does not work so well.
My system is designed as follows:
I create hundreds-thousands of IStrategy instances, each representing my trading strategy with different parameters.
I create one instance of ITesterClient
I create one TaskManager, which for each IStrategy creates a single Thread. In each thread, IStrategy is passed to ITesterClient. When ITesterClient finishes its job, thread stops and different thread is run instead. Task Manager ensures maximal number of Threads, which are running simultaneously.
When one thread is finished, it persist the test results and next thread starts running - this works as expected.
The whole process (testing all strategies) requires 10 hours or more to finish (even on my computational server). What is bothering me is one error, which occurs sometimes after 1 hour of computing, sometimes after 8 hours, but this error causes all running strategies to stop running.
Here are some relevant parts of my log file in time, when this happens. Unfortunately, no detailed stack trace is logged for the following error.
2017-10-25 14:04:28.571 INFO cz.jlochman.lochness.core.jforex.JForexSimpleReportClient - Strategy started: 9
2017-10-25 14:06:55.896 ERROR com.dukascopy.dds4.transport.client.netty.TransportClientSession - [NettyTransportClient-2] Request send failed because of SSLException: SSLEngine closed already
2017-10-25 14:06:55.901 ERROR com.dukascopy.dds4.transport.client.netty.TransportClientSession - [NettyTransportClient-2] Request send failed because of SSLException: SSLEngine closed already
2017-10-25 14:06:56.009 ERROR com.dukascopy.dds4.transport.client.netty.ClientConnector - Secondary session ping failed: javax.net.ssl.SSLException: SSLEngine closed already, timeout: 10000, synchRequestId: 20
2017-10-25 14:06:56.009 ERROR com.dukascopy.dds4.transport.client.netty.ClientConnector - Primary session ping failed: javax.net.ssl.SSLException: SSLEngine closed already, timeout: 10000, synchRequestId: 19
2017-10-25 14:06:56.107 INFO com.dukascopy.dds4.transport.client.netty.ClientConnector - Disconnect task in queue, reason [CONNECTION_PROBLEM], comments [Unknown connection problem], server address [d-ja-gva-91-191.dukascopy.com/194.8.15.191:10443], transport name [NettyTransportClient-2]
I have tried reconnecting my ITesterClient after certain amount of time, but the problem persisted. This happens on every machine I am using for my computations. They differ in operating system (Debian or OSX) and in internet connection, which they do not share.
Does anyone has an idea, why this is happening?
|
