|Posted on Wednesday, June 27, 2012 - 8:01 pm: |
this is driving me nuts...
i can submit jobs to gaussian and it runs without problem...BUT when i choose "Save Checkpoint" in the Advanced setting, I get the error below:
Entering Gaussian System, Link 0=/share/apps/g03/g03
/share/apps/g03/l1.exe /scratch/webmo-31011/79/Gau-17619.inp -scrdir=/scratch/webmo-31011/79/
buffer allocation failed in ntrext1.: Cannot allocate memory
Entering Link 1 = /share/apps/g03/l1.exe PID= 17620.
Is this something related to the checkpoint file?
Thanks for the help!
Post Number: 266
|Posted on Wednesday, June 27, 2012 - 9:08 pm: |
I am almost sure this is NOT a WebMO issue, per se. But I have never seen this particular issue before. You might try running a job on the command line to see if the same issue occurs.
|Posted on Sunday, July 01, 2012 - 12:17 am: |
i found out that this issue/bug is specific to g03 Rev E.01---the version of g03 that i'm using.
i solved it by having the run_gaussian.cgi change to the GAUSS_SCRDIR directory before running g03.
that went well, until run an NBO calculation. turns out that when i chdir to the scratch, the NBODATA files don't get copied. easy quick fix is to have the script copy the data in the scratch to the job home directory before the script recursively delete the scratch directory.
this works for me, but you might have a better/more elegant method...i'm all ears for that.
Post Number: 272
|Posted on Sunday, July 01, 2012 - 1:19 pm: |
Actually, that is going to be the best solution. It may not be elegant, but it's functional.