error message for mpp run of molpro

[Wheeler, Dr M.D.]

Dear All,
When I submit a job to use multiple smp nodes on our cluster

molpro -n6 filename.com

where filename.com is

***,Ar_benz3;
 memory,150,m;

                                                                                 !Define Z-matrix here
 geometry ={angstrom;
  q1
  q2   q1 rqq
  q3   q2 rqq       q1 90.000
  Ar   q2 rim       q1 theta       q3 phi
  c1   q2 rcc       q1 ccq         q3 0.0
  c2   q2 rcc       q1 ccq         q3 60.0
  c3   q2 rcc       q1 ccq         q3 120.0
  c4   q2 rcc       q1 ccq         q3 180.0
  c5   q2 rcc       q1 ccq         q3 240.0
  c6   q2 rcc       q1 ccq         q3 300.0
  h1   c1 rch       c2 120.0       c3 180.0
  h2   c2 rch       c3 120.0       c4 180.0
  h3   c3 rch       c4 120.0       c5 180.0
  h4   c4 rch       c5 120.0       c6 180.0
  h5   c5 rch       c6 120.0       c1 180.0
  h6   c6 rch       c1 120.0       c2 180.0
  }

                                                                                 !Variables for geometry set here
 rqq=  10.000000
 rim=  3.50
 theta=0.0
 phi = 90.0
 rcc=  1.2000
 ccq=  90.0
 rch=  1.08900

                                                                                 !do mp2 calculation
 basis= avdz;
 hf; save 2100.2
 mp2;
 mp4;
 ccsd(t);

 basis= avtz;
 hf; start 2100.2; save 2101.2;
 mp2;
 mp4;
 ccsd(t);

 ---;

The avdz part of the calculation runs smoothly over 6 processors.  the first few lines of the output are:
ARMCI configured for 3 cluster nodes

 MPP nodes  nproc
 kaitain       2
 giediprime    2
 caladan       2
 ga_uses_ma=false, calling ma_init with nominal heap. Any -G option will be ignored.

 Primary working directories:    /scratchdisk1:/scratchdisk2
 Secondary working directories:  /scratchdisk2:/scratchdisk1

 blaslib=atlas

 MPP tuning parameters: Latency=  1000 Microseconds,   Broadcast speed=   20 MB/sec
 default implementation of scratch files=ga

 ***,
 Including file /usr/local/lib/molpro-mpp-Linux-i686-i4-2002.6/molproi.rc


However, the avtz part of the calculation fails before the SCF stage giving the output at the end of the file:

 Eigenvalues of metric

         1 0.203E-05 0.271E-05 0.588E-05 0.290E-04 0.557E-04 0.832E-04 0.245E-03 0.308E-03
         2 0.164E-06 0.575E-06 0.885E-06 0.153E-05 0.108E-04 0.158E-04 0.327E-04 0.418E-04
         3 0.153E-05 0.108E-04 0.158E-04 0.940E-04 0.180E-03 0.263E-03 0.336E-03 0.386E-03
         4 0.187E-06 0.669E-06 0.203E-05 0.271E-05 0.588E-05 0.290E-04 0.323E-04 0.628E-04


 Contracted 2-electron integrals neglected if value below      1.0D-13
 AO integral compression algorithm  1   Integral accuracy      1.0D-13

     7080.772 MB (compressed) written to integral file ( 47.6%)

     Node minimum: 1109.918 MB, node maximum: 1300.496 MB


 NUMBER OF SORTED TWO-ELECTRON INTEGRALS:  244144104.     BUFFER LENGTH:  32768
 NUMBER OF SEGMENTS:   8  SEGMENT LENGTH:   31992480      RECORD LENGTH: 524288

 Memory used in sort:      32.55 MW
arrakis:/martynw/$ tail /home/mattd/laser2/caladan/ar-benz3.out
 Creating: host=kaitain, user=mattd,
           file=/usr/local/lib/molpro-mpp-Linux-i686-i4-2002.6/molprop_2002_6_athlon_tcgmsg.exe, port=33093
 Creating: host=giediprime, user=mattd,
           file=/usr/local/lib/molpro-mpp-Linux-i686-i4-2002.6/molprop_2002_6_athlon_tcgmsg.exe, port=33095
 Creating: host=caladan, user=mattd,
           file=/usr/local/lib/molpro-mpp-Linux-i686-i4-2002.6/molprop_2002_6_athlon_tcgmsg.exe, port=33097

 GA ERROR fehler

 GA ERROR


Does anyone have any idea what is causing this??  I am using the latest .rpm version of molpro2002.6 for athlon.

Cheers,
Martyn

----------------------------------------------
Dr. Martyn D. Wheeler
Department of Chemistry
University of Leicester
University Road
Leicester, LE1 7RH, UK.
Tel (office): +44 (0)116 252 3985
Tel (lab):    +44 (0)116 252 2115
Fax:          +44 (0)116 252 3789
Email:        martyn.wheeler at le.ac.uk
http://www.le.ac.uk/chemistry/staff/mdw10.html