CCL:G: Failure Restarting Gaussian 09 Jobs

 Sent to CCL by: Cory Pye [cpye\a/ap.smu.ca]
 Greetings from down the road.
 I think the problem comes from your use of ctrl-C.
 If you do a hard interrupt on a Gaussian job, it sometimes crashes in the
 middle of writing to the checkpoint file, or before it has a chance to. In
 other cases, the queueing system kills a job after say 48:00:00 hours, but the
 job hasn't yet finished, or it is in the middle of writing to the checkpoint
 file.
 It should work if you let the Gaussian program die on its own, as then it will
 write what it needs to the checkpoint file before error termination.
 -Cory
 On Wed, 20 Nov 2013, Douglas B. Staple dstaple^_^dal.ca wrote:
 >
 > Sent to CCL by: "Douglas B. Staple" [dstaple[*]dal.ca]
 > Hi Andrew,
 >
 > Yes, I can find the checkpoint file, but apparently Gaussian can't.
 > Good idea to try an absolute path, I didn't know that was possible.  I
 > tried:
 > %Chk=/home/dstaple/temp/benzene
 > and
 > %Chk=/home/dstaple/temp/benzene.chk
 >
 > Unfortunately both gave the same error as before.
 >
 > Thanks,
 >
 > Doug
 >
 >
 >
 >
 > --
 > Douglas B. Staple, Dr. rer. nat.
 > Department of Physics and Atmospheric Science
 > Dalhousie University
 > Halifax NS  B3H 4R2
 > Canada
 >
 >
 >
 > On 2013-11-20 1:03 PM, Andrew Yeung wrote:
 > > Hi Douglas
 > >
 > > Can Gaussian find the checkpoint file? Can you?
 > >
 > > How about using an explicit reference to the checkpoint file rather
 > > than an implicit reference?
 > >
 > > My checkpoint files have an extension of .chk. I'm not sure if
 > > Gaussian expects it.
 > >
 > > Andrew Yeung
 > > Donald J. Darensbourg Research Group
 > > Department of Chemistry, Texas A&M University
 > > 3255 TAMU
 > > College Station, TX 77843-3255
 > >
 > > Tel: 979 845 4837
 > > Fax: 979 845 0158
 > >
 > > On 2013-11-20 10:11 AM, Douglas B. Staple dstaple^^^dal.ca wrote:
 > >> Sent to CCL by: "Douglas B. Staple" [dstaple~~dal.ca]
 > >> For some reason I can't get Gaussian 09 to restart jobs that are
 > >> prematurely terminated.  Here is a test job:
 > >> http://myweb.dal.ca/~dstaple/benzene.gjf
 > >> Obviously the real jobs that are causing me issues are larger than
 > >> benzene, but the errors are exactly the same.
 > >>
 > >> I terminate the geometry optimization by hitting ctrl+C and then
 try
 > >> to restart the job using a new benzene.gjf file with the following
 form:
 > >> %NProcShared=8
 > >> %Mem=10GB
 > >> %Chk=benzene
 > >> # PBEPBE/AUG-cc-pVDZ SCF(NoVarAcc,Fermi,XQC)
 > >> Integral(Grid=UltraFine,Acc2E=12) Opt=Restart
 > >>
 > >> I've tried various modifications to the route line, such as:
 > >> # SCF(NoVarAcc,Fermi,XQC) Integral(Grid=UltraFine,Acc2E=12)
 Opt=Restart
 > >> or:
 > >> # PBEPBE/AUG-cc-pVDZ SCF(NoVarAcc,Fermi,XQC)
 > >> Integral(Grid=UltraFine,Acc2E=12) Opt Geom=AllCheck Guess=Read
 > >>
 > >> Each time, I get output that contains the error "FileIO
 operation on
 > >> non-existent file", ultimately terminating in a segmentation
 > >> violation.  I uploaded typical output (stdout and stderr) here:
 > >> http://myweb.dal.ca/~dstaple/benzene.e
 > >> http://myweb.dal.ca/~dstaple/benzene.o
 > >>
 > >> To make matters worse, the "real" jobs I'm interested in
 restarting
 > >> are multi-part jobs consisting of geometry optimization plus
 polarity
 > >> calculation, i.e. more like:
 > >> http://myweb.dal.ca/~dstaple/benzene_multipart.gjf
 > >> I know this changes things, but until I can get a geometry
 > >> optimization to restart it seems senseless to look at the
 multi-part
 > >> jobs.
 > >>
 > >> What am I doing wrong?
 > >>
 > >> Thanks,
 > >>
 > >> Douglas Staple
 > >> dstaple]*[dal.ca>
 >
    *************    !  Dr. Cory C. Pye
  *****************  !  Associate Professor
 ***   **    **  **  !  Theoretical and Computational Chemistry
 **   *  ****        !  Department of Chemistry, Saint Mary's University
 **      *  *        !  923 Robie Street, Halifax, NS B3H 3C3
 **      *  *        !  cpye- -crux.stmarys.ca   http://apwww.stmarys.ca/~cpye
 ***     *  *    **  !  Ph: (902)-420-5654  FAX:(902)-496-8104
  *****************  !
    *************    !  Les Hartree-Focks (Apologies to Montreal Canadien Fans)