Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[ptp-user] PTP debug limitation

Hi,

I'm not sure if this is a bug or a configuration problem (and I haven't seen a similar listing in the archives).

I have set up a new PTP environment for my graduate students that is being used to develop and debug on a remote Linux cluster (a frontend node and 16 backend compute nodes). The cluster sits behind a login server so SSH tunneling is being used for all connections. For the most part, everything is working. Projects can be created, built and run. The problem is with the PTP Debug perspective. I am able to debug applications properly, but only up to a limit of 5 processes (I am using one per node). Any attempt to use a larger process count causes the session to hang and then timeout after the "initialising debug processes (6)" box appears. PTP generates an error box saying "Error: start debugger request error". Eclipse provides the following error:

Error
Thu Oct 22 09:35:39 EDT 2009
Time out for this request.

org.eclipse.core.runtime.CoreException: Time out for this request.
at org.eclipse.ptp.debug.internal.core.PSession.connectToDebugger(PSession.java:153) at org.eclipse.ptp.launch.ParallelLaunchConfigurationDelegate$1$1.run(ParallelLaunchConfigurationDelegate.java:191) at org.eclipse.jface.operation.ModalContext$ModalContextThread.run(ModalContext.java:121)

There are three additional points to note. First, "running" the larger process counts is not a problem. I can run 17 processes either from the command line or from the PTP runtime perspective, using essentially the same run/debug configuration. Second, there does not appear to be any problem with the backend nodes themselves. In other words, I can modify the mpi_hosts files to include any group of 5 nodes (eg, frontend and nodes 1-4, frontend and nodes 6-9). This works just fine. I just can't add a 6th process to the debug configuration. Third, I have set up both PTP 2.1 and PTP 3.0. The result is the same in both versions.

Any ideas?

Cheers,

Todd

--
___________________________________

Todd Eavis, PhD
Associate Professor
Concordia Research Chair, Tier II
Department of Computer Science
Concordia University, Montreal
email: eavis@xxxxxxxxxxxxxxx
web: http://www.cs.concordia.ca/~eavis
___________________________________



Back to the top