From chemistry-request@server.ccl.net Thu Jun 6 03:35:33 2002 Received: from ribotargets.com ([194.129.39.11]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g567ZWL19253 for ; Thu, 6 Jun 2002 03:35:32 -0400 Received: from szilva (helo=localhost) by echo.ribotargets.com with local-esmtp (Exim 3.13 #1) id 17FrkJ-0000Gp-00; Thu, 06 Jun 2002 07:31:47 +0000 Date: Thu, 6 Jun 2002 07:31:47 +0000 (GMT) From: Szilveszter Juhos To: "Peterson,Brian K." cc: chemistry@ccl.net Subject: Re: CCL:EIDE vs SCSI & backup in Linux clusters In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII > current perspectives on disk solutions for Intel based Linux clusters? IMHO it is better to have a central RAID NFS server and good network connections than managing individual HDs on the cluster. (You do not need a HD at all on a cluster node though in practice most people buy nodes with HD since it is cheaper or just because that is easily available from vendors. Booting can be solved using network cards with EPROM). For RAID arrays I would recommend SCSI with a good controller and fine tuned kernel (summary here: http://server.ccl.net/cgi-bin/ccl/message.cgi?2002+05+07+009 ). Than backup becames easier since you have to backup only the RAID array not the individual nodes. Hope it helps: Szilva From chemistry-request@server.ccl.net Thu Jun 6 03:47:32 2002 Received: from oc30.uni-paderborn.de ([131.234.240.90]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g567lVL19675 for ; Thu, 6 Jun 2002 03:47:31 -0400 Received: (from ib@localhost) by oc30.uni-paderborn.de (SGI-8.9.3/8.9.3) id NAA73695; Thu, 6 Jun 2002 13:46:00 +0200 (CEST) Date: Thu, 6 Jun 2002 13:46:00 +0200 (CEST) Message-Id: <200206061146.NAA73695@oc30.uni-paderborn.de> X-Authentication-Warning: oc30.uni-paderborn.de: ib set sender to ib@oc30.uni-paderborn.de using -f From: Ingo Brunberg To: kckyung@inha.ac.kr CC: chemistry@ccl.net In-reply-to: <254361A45F4AD311B42900104B1343926656C9@inha.ac.kr> (message from =?euc-kr?B?sejC+bDm?= on Thu, 6 Jun 2002 01:22:54 +0900) Subject: Re: CCL:G98 : Printing isodensity surface References: <254361A45F4AD311B42900104B1343926656C9@inha.ac.kr> Dear Chan, you can use cube=density as a keyword to calculate the electron density at points in space distibuted on a regular cubic grid around a molecule. To find the points with an electron density of 0.001 au you have to do, what most graphics packages do: Calculate an isosurface at 0.001 au. There are a lot of programs which can calculate isosurfaces. A pure graphics application is OpenDX. From a chemists point of view programs like molden, gOpenMol or molekel come to mind. These are all free. An easy approach would be to use molden to import a Gaussian cube file, build the isosurface and save it as a vrml file. From that you can easily get the points on the surface. Regards, Ingo Brunberg > Dear Computational Chemists: > > I would like to print a set of points(x,y,z coordinates) on the > (molecular) surface of electron density 0.001 au of a molecule. Using this > points, I want to calculate electrostatic potential on this surface. But I > could not find an option to print this in the Gaussian 98 manual. Many > programs are available for the graphic representation of the surface. But I > need the cartesian coordinates of the surface. > > How can I do that? If not, do I need any other programs to generate such > coordinates? > > Thanks in advance. > > Chan. From chemistry-request@server.ccl.net Thu Jun 6 14:29:02 2002 Received: from front2.chartermi.net ([24.213.60.124]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56IT1L32158 for ; Thu, 6 Jun 2002 14:29:02 -0400 Received: from [24.247.80.10] (HELO quantumnt) by front2.chartermi.net (CommuniGate Pro SMTP 3.5.3) with SMTP id 8727014 for chemistry@ccl.net; Thu, 06 Jun 2002 14:29:02 -0400 Message-ID: <000d01c20d88$07f62380$0401a8c0@quantumnt> From: "Jim Kress" To: "CCL" Subject: Summary of CCL:guess=cards in G98W Date: Thu, 6 Jun 2002 14:29:01 -0400 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4807.1700 X-MIMEOLE: Produced By Microsoft MimeOLE V5.50.4807.1700 There was only one response. It was from Doug Fox of Gaussian. His response was: --------------------------------------------- There may be a buffer limitation. Some of the limitations needed to support Windows 3.1 may have survived. How big is the actual input file with all the MO's, over 512MB? On the workaround side, the SCF only cares about occupied orbitals as any set of orthogonal virtual orbitals will work. Do you actually have 281 occupied orbitals? Otherwise just truncate the list to the occupieds and you should be set. ------------------------------------------------- The truncation worked. Jim From chemistry-request@server.ccl.net Thu Jun 6 16:23:21 2002 Received: from mail.isg.siue.edu ([146.163.5.4]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56KNLL02081 for ; Thu, 6 Jun 2002 16:23:21 -0400 Received: from WEBSHIELD1.isg.siue.edu (webshield1.isg.siue.edu [146.163.5.149]) by mail.isg.siue.edu (8.9.1/8.9.1) with SMTP id PAA28195 for ; Thu, 6 Jun 2002 15:23:20 -0500 (CDT) Received: FROM mail.isg.siue.edu BY WEBSHIELD1.isg.siue.edu ; Thu Jun 06 15:23:19 2002 -0500 Received: from cougar.isg.siue.edu (cougar [146.163.5.29]) by mail.isg.siue.edu (8.9.1/8.9.1) with ESMTP id PAA28146 for ; Thu, 6 Jun 2002 15:23:13 -0500 (CDT) Received: from localhost (corrice@localhost) by cougar.isg.siue.edu (8.9.1/8.9.1) with ESMTP id PAA06074 for ; Thu, 6 Jun 2002 15:23:12 -0500 (CDT) Date: Thu, 6 Jun 2002 15:23:12 -0500 (CDT) From: COREY ALLEN RICE To: chemistry@ccl.net Subject: NiF CI calculation Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII TO WHOM THIS MAY CONCERN: I WAS WONDERING WHAT AN ERROR ON G98 MEANT. FAILURE BECAUSE OF LNK1E VIA I502.EXE. CAN ANYONE TELL ME WHAT THIS MEANS? Corey A Rice Southern Illinois University at Edwardsville Chemistry Department corrice@siue.edu http://www.siue.edu/~corrice From chemistry-request@server.ccl.net Thu Jun 6 16:40:00 2002 Received: from mail-a.bcc.ac.uk ([144.82.100.21]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56KdxL02383 for ; Thu, 6 Jun 2002 16:39:59 -0400 Received: from pop-c.ucl.ac.uk by mail-a.bcc.ac.uk with SMTP (Mailer) with ESMTP; Thu, 6 Jun 2002 21:38:20 +0100 Received: from ucl.ac.uk (chlorine.chem.ucl.ac.uk [128.40.76.17]) by pop-c.ucl.ac.uk (8.10.2+Sun/8.10.2) with ESMTP id g56Kc9o09266 for ; Thu, 6 Jun 2002 21:38:09 +0100 (BST) Sender: mb@ucl.ac.uk Message-ID: <3CFFC665.3E180DA@ucl.ac.uk> Date: Thu, 06 Jun 2002 21:30:29 +0100 From: Michael Brunsteiner Organization: University College London X-Mailer: Mozilla 4.76C-SGI [en] (X11; I; IRIX64 6.5 IP30) X-Accept-Language: en MIME-Version: 1.0 To: chemistry@ccl.net Subject: MD with net charge and Ewald ... Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-UCL-MailScanner: Found to be clean dear all, I plan do do a couple of classical MD simulations of systems with a net charge (aqueous ionic solution above a surface). When using a conventional Ewald summation algorithm the so called self term (which is usually included in standard implementations of Ewald) provides in such cases an effective background charge that takes care of the charge neutralisation more or less automatically. My questions: Is this (entirely) true ? What are possible artefacts introduced by using a charged system Any methodological problems I might encounter ? Any other known problems/pitfalls I might face with charged systems ? Any hints and/or references appreciated! cheers! Michael ========================================================================== According to convention there is a sweet and a bitter, a hot and a cold, and according to convention, there is an order. In truth, there are atoms and a void. -- Democritus, 400 B.C. -------------------------------------------------------------------------- Michael Brunsteiner Centre for Theoretical and Computational Chemistry University College London mailto:m.brunsteiner@ucl.ac.uk http://www.ucl.ac.uk/~uccambr/ -------------------------------------------------------------------------- From chemistry-request@server.ccl.net Thu Jun 6 18:06:16 2002 Received: from gandalf.cber.nih.gov ([128.231.52.5]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56M6FL03831 for ; Thu, 6 Jun 2002 18:06:15 -0400 Received: from localhost (rvenable@localhost) by gandalf.cber.nih.gov (980427.SGI.8.8.8/980728.SGI.AUTOCF) via ESMTP id SAA47689; Thu, 6 Jun 2002 18:05:54 -0400 (EDT) Date: Thu, 6 Jun 2002 18:05:54 -0400 From: Rick Venable To: Szilveszter Juhos cc: "Peterson,Brian K." , chemistry@ccl.net Subject: Re: CCL:EIDE vs SCSI & backup in Linux clusters In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII > > current perspectives on disk solutions for Intel based Linux clusters? On Thu, 6 Jun 2002, Szilveszter Juhos wrote: > IMHO it is better to have a central RAID NFS server and good network > connections than managing individual HDs on the cluster. (You do not need > a HD at all on a cluster node though in practice most people buy nodes > with HD since it is cheaper or just because that is easily available from > vendors. Booting can be solved using network cards with EPROM). For RAID > arrays I would recommend SCSI with a good controller and fine tuned kernel > (summary here: > http://server.ccl.net/cgi-bin/ccl/message.cgi?2002+05+07+009 ). Than > backup becames easier since you have to backup only the RAID array not the > individual nodes. However, for large scratch files such as one might need for QM calculations, a local disk would be preferred (faster access) over an NFS mounted RAID server, which has network overhead. I do agree that one doesn't want to use the cluster nodes for any other type of storage. =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= Rick Venable 29/500 FDA/CBER/OVRR Biophysics Lab 1401 Rockville Pike HFM-419 Rockville, MD 20852-1448 U.S.A. (301) 496-1905 Rick_Venable@nih.gov ALT email: rvenable@speakeasy.org =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= From chemistry-request@server.ccl.net Thu Jun 6 02:55:57 2002 Received: from jubilantbiosys.com ([202.144.86.69]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g566tsL17706 for ; Thu, 6 Jun 2002 02:55:55 -0400 Received: from parthiban ( [192.168.1.56] (may be forged)) (authenticated) by jubilantbiosys.com (8.11.6/8.11.6) with ESMTP id g566vit12511 for ; Thu, 6 Jun 2002 12:27:44 +0530 Message-ID: <07e601c20d26$fba973e0$3801a8c0@parthiban> From: "Parthiban Srinivasan" To: Subject: Freeware for drug design Date: Thu, 6 Jun 2002 12:24:19 +0530 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_07E3_01C20D55.15049BC0" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6600 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 This is a multi-part message in MIME format. ------=_NextPart_000_07E3_01C20D55.15049BC0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Dear Friends: I am looking for downloadable programs (trivial to sophisticated) = applicable in=20 various stages of drug discovery pipeline. Kindly send me the URL and/or = any other helpful info. As an useful prcatice, I will summarise the = replies. Thanks and Regards S Parthiban Jubilant Biosys ------=_NextPart_000_07E3_01C20D55.15049BC0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Dear Friends:
I am looking for downloadable programs = (trivial to=20 sophisticated) applicable in
various stages of=20 drug discovery pipeline. Kindly send me the URL and/or
any other helpful info. As an useful = prcatice, I=20 will summarise the replies.
Thanks and Regards
S Parthiban
Jubilant Biosys
 
------=_NextPart_000_07E3_01C20D55.15049BC0-- From chemistry-request@server.ccl.net Thu Jun 6 11:11:18 2002 Received: from hotmail.com ([64.4.19.236]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56FBIL28212 for ; Thu, 6 Jun 2002 11:11:18 -0400 Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Thu, 6 Jun 2002 08:11:13 -0700 Received: from 137.48.16.30 by lw12fd.law12.hotmail.msn.com with HTTP; Thu, 06 Jun 2002 15:11:13 GMT X-Originating-IP: [137.48.16.30] From: "GUANGPING LI" To: chemistry@ccl.net Subject: atomic density Date: Thu, 06 Jun 2002 10:11:13 -0500 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 06 Jun 2002 15:11:13.0666 (UTC) FILETIME=[65E0DE20:01C20D6C] Dear Friends: Could you please tell me how to calculate and get a contour map of the deformation electron densities of a plannar molecule? I was trying to produce such a contour map using the gaussian 98 and AIM (atoms in molecules) package. I have tried both ghost atom with basis set and real atom without basis set techniques, but the results were not good enough. Best Wishes, JIM _________________________________________________________________ Send and receive Hotmail on your mobile device: http://mobile.msn.com From chemistry-request@server.ccl.net Thu Jun 6 10:12:28 2002 Received: from rwcrmhc52.attbi.com ([216.148.227.88]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56ECSL27253 for ; Thu, 6 Jun 2002 10:12:28 -0400 Received: from daveg-lifebook.cachesoftware.com ([12.224.212.36]) by rwcrmhc52.attbi.com (InterMail vM.4.01.03.27 201-229-121-127-20010626) with ESMTP id <20020606141223.JYUK2751.rwcrmhc52.attbi.com@daveg-lifebook.cachesoftware.com>; Thu, 6 Jun 2002 14:12:23 +0000 Message-Id: <5.1.0.14.0.20020606070221.028ec010@pop3.norton.antivirus> X-Sender: (Unverified) X-Mailer: QUALCOMM Windows Eudora Version 5.1 Date: Thu, 06 Jun 2002 07:16:53 -0700 To: Roy Jensen From: David Gallagher Subject: Re: singlet-triplet crossings... Cc: chemistry@ccl.net In-Reply-To: Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="=====================_41443277==_.ALT" --=====================_41443277==_.ALT Content-Type: text/plain; charset="us-ascii"; format=flowed Hi Roy, Angalda and Bofill have developed a method for locating intersystem crossing geometries which is now incorporated into MOPAC 2002 and CAChe. The reference below may provide more insight, and there is additional information in the MOPAC manual at http://www.cachesoftware.com/techsupport/mopac/ J. M. Anglada and J. M. Bofill, "A Reduced Restricted-Quasi-Newton-Raphson Method for Locating and Optimizing Energy Crossing Points Between Two Potential Energy Surfaces", J. Comp. Chem., 1997, 18, 992-1003 (1997). Regards, David Gallagher At 11:13 AM 6/5/2002 -0700, Roy Jensen wrote: >All > >I would like to understand how states of different multiplicity are >coupled together AND how to determine the probability of surface-crossing. > >Coupling is, I believe, a spin-orbit interaction. CASSCF can apparently >optimize to a singlet-triplet crossing -- has anyone attempted this? >I am aware of the work by Landau and Zener in the 1930s on surface >crossing probabilities. > >Roy Jensen > > > > >-= This is automatically added to each message by mailing script =- >CHEMISTRY@ccl.net -- To Everybody | CHEMISTRY-REQUEST@ccl.net -- To Admins >MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH >CHEMISTRY-SEARCH@ccl.net -- archive search | Gopher: gopher.ccl.net 70 >Ftp: ftp.ccl.net | WWW: http://www.ccl.net/chemistry/ | Jan: jkl@ccl.net --=====================_41443277==_.ALT Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi Roy,

Angalda and Bofill have developed a method for locating intersystem crossing geometries which is now incorporated into MOPAC 2002 and CAChe. The reference below may provide more insight, and there is additional information in the MOPAC manual at http://www.cachesoftware.com/techsupport/mopac/

J. M. Anglada and J. M. Bofill, "A Reduced Restricted-Quasi-Newton-Raphson Method for Locating and Optimizing Energy Crossing Points Between Two Potential Energy Surfaces", J. Comp. Chem., 1997, 18, 992-1003 (1997).

Regards,
David Gallagher

At 11:13 AM 6/5/2002 -0700, Roy Jensen wrote:
All

I would like to understand how states of different multiplicity are
coupled together AND how to determine the probability of surface-crossing.

Coupling is, I believe, a spin-orbit interaction. CASSCF can apparently
optimize to a singlet-triplet crossing -- has anyone attempted=20 this?
I am aware of the work by Landau and Zener in the 1930s on surface
crossing probabilities.

Roy Jensen

<I will summarize to the List.>


-=3D This is automatically added to each message by mailing script =3D-
CHEMISTRY@ccl.net -- To Everybody  | CHEMISTRY-REQUEST@ccl.net -- To Admins
MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH
CHEMISTRY-SEARCH@ccl.net -- archive search    |    Gopher: gopher.ccl.net 70
Ftp: ftp.ccl.net |  WWW: http://ww= w.ccl.net/chemistry/  | Jan: jkl@ccl.net
--=====================_41443277==_.ALT-- From chemistry-request@server.ccl.net Thu Jun 6 04:21:17 2002 Received: from mail.kemi.dtu.dk (IDENT:root@[130.225.67.150]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g568LHL20969 for ; Thu, 6 Jun 2002 04:21:17 -0400 Received: from ouabain.ok.dtu.dk (tr4.ok.dtu.dk [192.38.70.74]) by mail.kemi.dtu.dk (8.11.6/8.11.4) with ESMTP id g568LGJ04911; Thu, 6 Jun 2002 10:21:17 +0200 Received: from malcolm by ouabain.ok.dtu.dk with local (Exim 3.12 #1 (Debian)) id 17FsWB-00011k-00; Thu, 06 Jun 2002 10:21:15 +0200 Date: Thu, 6 Jun 2002 10:21:15 +0200 To: chemistry@ccl.net Cc: "Peterson,Brian K." Subject: Re: CCL:EIDE vs SCSI & backup in Linux clusters Message-ID: <20020606102115.K301@tr4.ok.dtu.dk> Mail-Followup-To: chemistry@ccl.net, "Peterson,Brian K." References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: ; from PETERSBK@apci.com on Wed, Jun 05, 2002 at 04:37:02PM -0400 From: Malcolm Gillies On Wed, Jun 05, 2002 at 04:37:02PM -0400, Peterson,Brian K. wrote: > We are specifying a Linux Cluster to use for a variety of codes including > computational chemistry- both quantum and molecular simulations. I've > looked at the CCL archives to see what people have said about SCSI vs EIDE > drives in such a cluster and the results are a bit ambiguous. What are the > current perspectives on disk solutions for Intel based Linux clusters? Are > EIDE drives sufficient? Does one need to tune parameters to get good > performance? SCSI vs IDE can turn into a bit of religious issue, so you are likely to run into some conflicting opinions. The technology is also changing quickly, with various different generations of interface available on the market simultaneoulsy, which makes matters more confusing. The general opinion seems to be that with the more recent generations of IDE interfaces (UDMA 66 and up), IDE no longer has a systematic transfer rate disadvantage compared to SCSI, i.e. for both SCSI and IDE, interface transfer rates are well above the maximum physical transfer rates of the available drives. However, new drive technology usually appears first with a SCSI interface. For instance, I believe that 15 000 rpm drives are currently only available as SCSI. If you need the highest possible performance, then SCSI is the only option. On the other hand, SCSI drives tend to be larger, noisier and dissipate more heat. The remaining advantage of SCSI for me is the ability to swap drives easily, and to connect drives in an external enclosure. As far as I know, hot-swapping of IDE drives is currently impossible. While this is useful for file servers, or when configuring machines, it's less important for scratch disks on cluster machines. We have a cluster here with both SCSI-based and IDE-based nodes. If we buy new nodes in the future, I would recommend IDE, as the price advantage is quite convincing. If disk performance is a limiting factor, a RAID setup may also be solution. As for tuning parameters, this will depend on your hardware. For most chipsets, with a recent Linux kernel (2.4.*) it is no longer necessary. Ideally, you should test before you buy. With Intel x86-based hardware, it should be easy enough to find an alternative if a particular type of motherboard/chipset turns out to be problematic. The rule of thumb here is to stick with hardware that has been on the market for a while, say six months to a year. > Second question. What are you doing about backup of disks on your Linux > clusters? Do you just move output files to another machine which is backed > up? We don't back up anything on the compute nodes. The queuing system (OpenPBS) is responsible for moving output files back to the master node, which is backed up. cheers, Malcolm -- Malcolm Gillies Post-doctoral researcher Organic Chemistry, Dept Chemistry, Technical University of Denmark From chemistry-request@server.ccl.net Thu Jun 6 14:02:56 2002 Received: from mail.bancorp.ru (root@[217.107.81.59]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g56I2tL31669 for ; Thu, 6 Jun 2002 14:02:55 -0400 Received: from 217.107.75.31 (dialup31-kazan.telebit.ru [217.107.75.31] (may be forged)) by mail.bancorp.ru (8.11.3/8.11.3) with ESMTP id g56I2nc14187; Thu, 6 Jun 2002 22:02:50 +0400 Date: Thu, 6 Jun 2002 22:05:14 +0400 From: Gregory Shamov X-Mailer: The Bat! (v1.45) Educational Reply-To: Gregory Shamov X-Priority: 3 (Normal) Message-ID: <1671451917.20020606220514@bancorp.ru> To: Szilveszter Juhos CC: chemistry@ccl.net Subject: Re: CCL:EIDE vs SCSI & backup in Linux clusters In-reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello Szilveszter, Thursday, June 06, 2002, 11:31:47 AM, you wrote: >> current perspectives on disk solutions for Intel based Linux clusters? SJ> IMHO it is better to have a central RAID NFS server and good SJ> network connections than managing individual HDs on the cluster. For common parallel qchemistry packages (say, GAMESS) for diskless cluster you will need _very_ good connections for multigigabyte NFS disk I/O. I'm not sure that Ethernet (even Gigabit Ethernet) will be fast enough. SJ> (You do not need a HD at all on a cluster node though in practice SJ> most people buy nodes with HD since it is cheaper or just because SJ> that is easily available from vendors. Booting can be solved using SJ> network cards with EPROM). For RAID arrays I would recommend SCSI SJ> with a good controller and fine tuned kernel (summary here: SJ> http://server.ccl.net/cgi-bin/ccl/message.cgi?2002+05+07+009 ). SJ> Than backup becames easier since you have to backup only the RAID SJ> array not the individual nodes. -- Best regards, Gregory mailto:gas5x@bancorp.ru