From b_duke@lacebark.ntu.edu.au Wed Jul 24 00:16:00 1996 Received: from lacebark.ntu.edu.au for b_duke@lacebark.ntu.edu.au by www.ccl.net (8.7.5/950822.1) id XAA24897; Tue, 23 Jul 1996 23:33:28 -0400 (EDT) From: Received: by lacebark.ntu.edu.au (AIX 3.2/UCB 5.64/4.03) id AA13999; Thu, 25 Jul 1996 12:37:06 +1100 Message-Id: <9607250137.AA13999@lacebark.ntu.edu.au> Subject: G94: Splitting RWF. To: CHEMISTRY@www.ccl.net (chemistry) Date: Thu, 25 Jul 1996 12:37:05 +1100 (EETDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Hi folks, I am trying to work out how to split the RWF file between a local disk and a remote NFS disk. /scr/qchem/ is the local disk. /lacescr/ is the remote disk mount point. Each are 2Gb but since there is a bit on them I played save and used 230Mwords. 1. If I give 230 as the size of both files it straight away fails and tells me there is no file to extend. 2. If I give -1 for the local file as below, it fails with about 1.5Gb on the local disk and nothing on the remote one. ************************************************* Gaussian 94: DEC-AXP-OSF/1-G94RevC.3 26-Sep-1995 23-Jul-1996 ************************************************* %chk=/usr/users/b_duke/chk/ga2h2_dbm %RWF=/lacescr/,230,/scr/qchem/,-1 Default route: MaxDisk=250000000 ---------------------------------------------------------------------- # mp2 gen 5d scf=direct geom=checkpoint guess=read freq MaxDisk=450000 000 ---------------------------------------------------------------------- alpha-alpha T2 = 0.1696787838D-01 E2= -0.9422155339D-01 alpha-beta T2 = 0.1112765387D+00 E2= -0.4149900717D+00 beta-beta T2 = 0.1696787838D-01 E2= -0.9422155339D-01 ANorm= 0.1070145923D+01 E2= -0.6034331785D+00 EUMP2= -0.38482541950061D+04 PickT4: no shell combinations can fit! NKLS2p= 49 NKLS2= 49 MaxCom= 37 Error termination via Lnk1e in /usr/users/qchem/g94/l1111.exe. Job cpu time: 0 days 2 hours 46 minutes 56.7 seconds. File lengths (MBytes): RWF= 1600 Int= 0 D2E= 0 Chk= 3 Scr= 1 3. If I swap these round and give -1 for the remote disk and 230 for the local disk, it writes to the remote disk and then fails in a different way . ************************************************* Gaussian 94: DEC-AXP-OSF/1-G94RevC.3 26-Sep-1995 24-Jul-1996 ************************************************* %chk=/usr/users/b_duke/chk/ga2h2_dbm %RWF=/lacescr/,-1,/scr/qchem/,230 Default route: MaxDisk=250000000 ---------------------------------------------------------------------- # mp2 gen 5d scf=direct geom=checkpoint guess=read freq MaxDisk=450000 000 ---------------------------------------------------------------------- Range of M.O.s used for correlation: 19 138 NBasis= 138 NAE= 32 NBE= 32 NFC= 18 NFV= 0 NROrb= 120 NOA= 14 NOB= 14 NVA= 106 NVB= 106 Semi-Direct transformation. ModeAB= 4 MOrb= 32 LenV= 7488722 LASXX= 16071592 LTotXX= 16071592 LenRXX= 32641304 LTotAB= 16569712 MaxLAS= 20132544 LenRXY= 0 NonZer= 48712896 LenScr= 80558066 LnRSAI= 20132544 LnScr1= 37687538 MaxDsk= 450000000 Total= 171019452 SrtSym= T fsync failed in NtrExt1. 4. If I run the same job on an identical Dec alpha writing to just one 4Gb disk, the job works fine. The checkfile and com file were just copied across from the first machine. So, the job is OK. What is going on? How does the file splitting work? Why does it start writing on the file that I specified as -1, rather than the first file on the line. Obviously I want to write to the local file first and only extend to the NFS mounted disk if I have to. I would appreciate any help and advice on this question. Cheers, Brian. -- Associate Professor Brian Salter-Duke (Brian Duke) School of Mathematical and Physical Sciences, Northern Territory University, Darwin, NT 0909, Australia. Phone 08-89466702. Fax 08-89410460 e-mail: b_duke@lacebark.ntu.edu.au WWW http://lacebark.ntu.edu.au/chemistry