I am writing this email to ask for help in setting up a debugging environment for openMPI-based applications. Currently, I am struggling with a problem that is hampering me from continuing the work. The problem is very similar to what has been posted on the stackoverflow (See: Link). Its symptom is that the debugging process hangs and the parallel debugging perspective shows a “Operation in progress…” message. It is caused by an error, namely “Cannot connect to debugger”. Meanwhile, an error message “getaddrinfo: Undefined error: 0” also appear in the debugging console.
I did some search on the internet and I found an instruction for trouble shooting (See: Link). I already verified that:
Launch script could work: non-debug launch works fine
Correct number of “sdm” processes
Routing file can be created
But I could not find one “sdm” process with argument “—master”.
Currently, I use the following software components with indicated versions:
Parallel Tools Platform including PTP Parallel Language Development Tools, etc. (9.0.1.201509091505)
ptp-sdm-9.0.0 (org.eclipse.ptp.macosx_9.0.0.201506101404)
GNU gdb (7.10)
The OS I am using is OS X Yosemite (10.10.5). And I run application on the local machine.
Did anyone encounter the similar problem and happen to know how to fix it? I would appreciate any hint or help from you.
I've got the similar problem under Ubuntu 16.04.2, but just “Cannot connect to debugger”. My solution is to run Eclipse with superuser privileges, i.e. sudo ./eclipse. Hope it is useful.
I found something that might be the cause. I ran Eclipse from a shell terminal so that I could see more text/log output. Then under Eclipse -> Preference -> Parallel Tools -> Parallel Language Development Tools I enabled the Show output from commands, and this is what I see in the parent shell (not within Eclipse):
Cleanup: rm -f routing_file
submit-interactive-debug: ac5df46b-c4f2-438a-9565-9b26818e880d: perl /Users/<username>/.eclipsesettings/rms/MPICH2/start_debug_job.pl mpiexec -np 4 /Users/<username>/.eclipsesettings/sdm --port=51347 --host=localhost --debugger=gdb-mi
SDM Master: /Users/<username>/.eclipsesettings/sdm -master --port=51347 --host=localhost --debugger=gdb-mi
SDM Master: sdm [--debugger=value] [--debugger_path=path]
SDM Master: [--proxy=proxy]
SDM Master: [--host=host_name] [--port=port]
SDM Master: [--master]
SDM Master: [--server=rank]
SDM Master: [--generate_routes=list]
SDM Master: [--routing_file=path]
SDM Master: [--debug[=level]]
This suggests that one of the argument inputs to sdm is invalid, and indeed I see that it is receiving -master instead of --master flag.
I tried to open the perl script at /Users/<username>/.eclipsesettings/rms/MPICH2/start_debug_job.pl but it's correct there, and any changes made only appear within Eclipse's output console, not the parent shell I ran eclipse from. Somewhere sdm is being sent the wrong -master flag, and maybe that is part of the problem. Unfortunately I don't have an answer beyond that.
Update: I have submitted a bug report: https://bugs.eclipse.org/bugs/show_bug.cgi?id=521975
Related
I am trying to setup coverity in my system, while configuring the compiler I am getting the below error:
I have uninstall the setup and try to install freshly but no luck yet.
Any body has any idea?
This typically occurs when the compiler does not run in the environment you're using with cov-configure.
A quick and easy way to check would be to attempt to compile something trivial and simple - like a "Hello World" program in the same console window that you're running cov-configure from. Since Coverity does support that version of QNX, I expect the issue is an environment one.
I am using Valgrind to debug an OpenCV application. My application often stops with Segmentation Fault error. I thought this is related to memory, so I debug it using Valgrind in Eclipse. Unfortunately, when I run the valgrind profile in Eclipse, it returns : cannot connect to X server and valgrind stops executing my application.
According to this post, I have set DISPLAY variable to 0.0, but same behavior still happened.
I also tried installing xorg and openbox, but still nothing happened.
The application runs until before I called "cvNamedWindow(WINDOW_TITLE, CV_WINDOW_AUTOSIZE);". On executing the line, Valgrind outputs : cannot connect to X server and then stops.
Based on my previous experience with Valgrind, I expect it will execute the application, and exits when memory problem happened/application terminated, because that is what happened if I run the application normally (Segmentation Fault). Also, thats how I debug multithreading in in an application, previously.
Any help is appreciated. Thanks!
I tried running the application with valgrind, root privilege in terminal:
`sudo valgrind --tool=memcheck <path-to-application>/myapp`
and it works. Seems like the Eclipse plugin do not have root access to X server/display. The problem is solved
I'm using this amazing IPython notebook. I'm very interested into parallel computing right now and would like to use MPI with IPython (and MPI4py). But I can't start a cluster with
ipcluster start -n 4
on Windows7. I just get back "failed to create process". If I use the notebook and start a cluster in the "Clusters" register it's all working fine. But with cmd (even with admin rights) I just get this message. Same with all attempts of using MPI (MPICH2). All path vars are set. Maybe this problem has no connection to Python at all...
I can't say anything about IPython's parallel features, but if you're having problems with MPI in Windows in general, I would offer these suggestions. I've had quite a few issues in the past in trying to get MPI working in Windows. The most convenient method for me in the past has been to use an OpenMPI Windows binary http://www.open-mpi.org/software/ompi/v1.6/. These are now only available in previous releases. And even then, you might have to try more than one before you find one that works. I don't know why, but the latest didn't work on my machine. The release before that one did, however. After this, you have to call mpicc and mpiexec from the Microsoft Visual Studio Command Prompt or it won't work (without a lot of other stuff).
After you have verified that MPI is working, you can try installing mpi4py separately and see if that works. In my experience, sometimes this has worked fine and sometimes I've had to wrestle with configurations. You might just try your luck with an unofficial, prepackaged binary (for example, http://www.lfd.uci.edu/~gohlke/pythonlibs/).
Hope this helps!
I installed Valgrind on SUSE Linux SLES 11SP2 under my home directory and added it to the path. I then installed eclipse Juno for C/C++ Service Release 2 Build id: 20130225-0426 and installed the Valgrind plugin.
When I try to create a profile configuration with Valgrind I get an error message:
"[Valgrind Options]: Couldn't determine version of Valgrind", and though it allows me to make the configuration it prevents me from proceeding by greying out all buttons except for the close button.
I know I am missing something trivial like defining the location of the Valgrind executable in eclipse, but I was unable to find it, not in the project properties nor in the window preferences.
I will appreciate your help.
Ok, I figured it out.
I moved the Valgrind installation to a new shared location, and tried it out with a new computer (same architecture).
It did not affect the original computer I installed it on, but Valgrind had a problem with the new one, as indicated in the README (and I meanwhile forgot):
“Important! Do not move the valgrind installation into a place different from that specified by --prefix at build time. This will cause things to break in subtle ways, mostly when Valgrind handles fork/exec calls.”.
This caused Valgrind to break when run standalone with the message:
“valgrind: failed to start tool 'memcheck' for platform 'amd64-linux': No such file or directory”,
however when running Valgrind from eclipse it probably translated it to the message:
"[Valgrind Options]: Couldn't determine version of Valgrind".
In short: Do not move the Valgrind installation but re-install it. And this message from eclipse indicates that something bad happened with the installed Valgrind and not that it cannot find it.
I am getting the same error but its not caused by the explanation you have given.
You can get this error if you are using valgrind on mac os x because valgrind is usually not installed in the directory eclipse is looking for. In order to fix the issue on mac, you need to make a symbolic link to the valgrind location by
typing ln -s valgrind_location /usr/bin/valgrind. Note this differs from what are in the directions on eclipse's site http://wiki.eclipse.org/Linux_Tools_Project/Valgrind/User_Guide. For some reason they have the call backwards on their site. Once I made this correction valgrind worked correctly from within Eclipse.
So on macs at least the error that it can't determine the version of valgrind at least matches up with eclipse not being able to find the file.
I have a CVS server which is known to be ok (works from other machines).
I am trying to set up Eclipse to connect from an Ubuntu box.
The following command-line command succeeds:
cvs -d ':extssh:myuser#myhost/path/to/repository' checkout myrepository
Yet when I do Eclipse 'Add CVS Repository' it hangs on this, using both extssh and pserver protocols. extssh using port 22.
(There's no error message, it just hangs. Regardless whether 'Validate connection on finish' is on or off.)
I verified that all of the settings are ok. Port 22 is not blocked.
I double-checked Preferences under 'General>Network Connections' and 'Team>CVS'
I do not think it is an issue with keyless ssh either.
This is on an Ubuntu box, but these exact same Eclipse CVS client settings succeed from a Mac box.
(The Ubuntu box uses 9.10, and Eclipse is installed as EasyEclipse 1.3.1, installed as user, not root.
Plugin org.eclipse.team.* versions are:
(Ubuntu)
org.eclipse.team.cvs.ssh,.ssh2 3.2.100
org.eclipse.team.ui,.core,.cvs.core,.cvs.ui 3.3.r33x_2007...
(Mac)
org.eclipse.team.cvs.ssh2 3.2.300
org.eclipse.team.ui,.cvs.core,.cvs.ui 3.5.100.l20100527-0800
(EasyEclipse prevents me from upgrading the Ubuntu plugins, it insists these 4-year-old ones are the latest. Maybe an argument against using EasyEclipse.)
How to troubleshoot? How to trace what is actually happening inside Eclipse?
(as 'cvs -t -t' would give)
(As a sidebar, Eclipse should actually be printing a proper error message.
I've checked every appnote and userguide I can find with Google.)
Consensus was that EasyEclipse prevents user from upgrading the (4-year-old) Ubuntu plugins, insists that 3.3.x are the latest available, which cause this hang.
As of 5/2011, EasyEclipse is retired unless a new maintainer steps up.