RM consumes processor cycles indefinitely if internet connected during startup

If my PC is connected to the internet, RM consumes a lot of processor cycles (about 1 full core's worth) continuously while idling. Even if I just start RM (with all extensions disabled) and let it stand on the welcome screen for many hours, RM will continue to consume 1 full core's worth of processor cycles. So much so that the processor's fan has to work quite hard.

But this does not happen if there was no internet connection when RM was started. There even seems to be a correlation between processor time consumed, and internet connection speed, even though the connection itself does not seem busy.

Moreover, if the connection is broken after RM is started, then RM continues to consume the processor time at the same rate, indefinitely.

My only recourse currently is to disconnect from the internet every time I want to start RM, and reconnect afterwards. Otherwise everything runs a little slower.

Find more posts tagged with

AI Studio

Accepted answers

All comments

MariusHelf

Which operating system are you using? Do you have a firewall or a connection view that allows you to find out which servers are accessed by RapidMiner/Java? If you start RapidMiner from the console, can you see any unnormal entries in the output? Unfortunately, we are not able to reproduce the problem.

Best regards,
Marius

tennenrishin

Windows 7. I don't know how to determine which servers are accessed by RapidMiner, and there are no unusual messages in the RM console.

A correction:
[quote=tennenrishin]There even seems to be a correlation between processor time consumed, and internet connection speed[/quote]
Actually it seems to happen only on one (of three) of my internet connections, and not on others, regardless of bandwidth. The connection in question is a mobile broadband connection (3G) using a HAUWEI USB dongle that establishes a remote-NDIS-based connection to the internet. Other Java applications don't have this problem on this connection, so it is unlikely to be happening within the Java API.

If it isn't a common problem then perhaps it isn't worth spending much time on. I'll just use other connections when running RM.

Thank you for having looked into it
Isak