| T O P I C R E V I E W |
| dvsmith |
Posted - Oct 03 2008 : 4:06:27 PM
Any ideas why bxhreorient would be failing on this data set?
SocReward.03
20081001_34990
run005_06
bxhreorient --orientation=LAS run005_06.bxh reoriented.bxh
Reading data...
Reorienting data...
bxhreorient: Error writing to data file 'reoriented.dat'
bxhreorient: failed.
I can't see any obvious problems in the .bxh header, but I suspect something has to be screwed up there. The data look fine (aside from lots of motion) when I view it in MATLAB.
Thanks,
David
|
| 7 L A T E S T R E P L I E S (Newest First) |
| syam.gadde |
Posted - Oct 04 2008 : 10:26:22 AM
That particular error should only happen when the file has been created/opened successfully, so permissions should not be an issue. If it happens again, perhaps you can record on what node these errors happen and maybe the size of the data (it's possible there is a memory issue). |
| dvsmith |
Posted - Oct 03 2008 : 5:56:07 PM
I've written scripts to use the Linux-based BXH tools via the cluster, and I've written scripts to use the Windows-based BXH tools via MATLAB. My cluster one failed last night, on just that one run. Now it magically works on the cluster, which is weird.
I don't see why executing a system command (with ! or "system") in MATLAB would fail, but I guess that's possible.
Thanks,
David
|
| petty |
Posted - Oct 03 2008 : 5:30:44 PM
wait, are you running reorient through matlab?
you can run it straight through the command line in the linux environment, which i've never had a problem with.
if it is in fact through matlab, there are always memory type issues that come up unexpectedly. sometimes large displays work, sometimes they crash etc. |
| dvsmith |
Posted - Oct 03 2008 : 5:25:48 PM
Why would it be an ownership issue within my experiment folder that is specific to a single run? Weird problems with the cluster? I guess it wouldn't be the first time the cluster behaved strangely...
Anyway, I've ran it 4 times in MATLAB with the same exact error:
bxhreorient --orientation=LAS run005_06.bxh reoriented.bxh
Reading data...
Reorienting data...
bxhreorient: Error writing to data file 'reoriented.dat'
bxhreorient: failed.
If the same file is there already, it throws a different error. So it's not that.
Oddly enough, when I run the same exact code from last night on the cluster, it works now. Very strange. Should we just chalk this up to the cluster being buggy? Do you think it could be a memory/RAM issue? No one else was on there when I was running this last night. I don't understand why something would fail to work in one instance and then magically work in the next instance, especially since it's the same code. It appears to be completely random.
Thanks,
David
|
| petty |
Posted - Oct 03 2008 : 4:30:46 PM
well if its any consolation i just entered the same code for your same subject's run, but wrote it to my home drive and didn't receive any errors.
a potential directory ownership issue?
also, if there already is a file of the same name i think it will fail. |
| dvsmith |
Posted - Oct 03 2008 : 4:27:46 PM
nope. we've got 31.2 GBs left on that study. |
| petty |
Posted - Oct 03 2008 : 4:25:00 PM
could there be a space issue?
(ie: not enough space left in the study)
try to write it out to your home directory or something and see what happens
|
