  COCOA Beowulf Cluster FAQ

  Anirudh Modi anirudh-modi@psu.edu
  Version 0.1.2 1999 6 2
  :   , yooseong@kldp.org
  : 2000 3 17

    Beowulf ǻ͸    ʿ 
  Ѵ.         ַ ü
  ġ   ߰ ִ.  HOWTO diskless Ŭ̾Ʈκа
  Ŭ͸   Ű ϴ  ִ.  HOWTO
    5.2 Ŀ 2.0.x  ϰ ִ.
  ______________________________________________________________________

  

  1. 
  2. COCOAۿ  󼼻
     2.1 ϵ ġ
     2.2 Ʈ ġ

  ______________________________________________________________________

  1.  

      COCOAΰ?
        COCOA COst effective COmputing Array ڷμ Beowulf η
        ǻԴϴ. Beowulf ó   ִ 
        ǻ   ִ. ̴ ϳ  ̴ ī峪
        ٸ  Ʈũ  ϳ ̻ Ŭ̾Ʈ 
        ̷ִ. ȭǾ ִ ǥ ̴ Ϳ ġ
         ũž PC   ϵ ̿   ִ.
        ȭ   ϵ      ִ.
        <http://cocoa.ihpca.psu.edu/>

     COCOA µ  ϵ ʿѰ?
        Dell <http://www.dell.com>  26 WS-410
        ũ̼ǵε, Ҵ  .

        1. Dual 400 MHz Intel Pentium II Processors w/512K L2 cache

        2. 512 MB SDRAM

        3. 4 GB UW-SCSI2 Disk

        4. 3COM 3c509B Fast Ethernet adapter (100 Mbits/sec)

        5. 32x SCSI CD-ROM drive

        6. 1.44 MB floppy drive

        7. Cables

        ߰  뵵 Ǿ:

        1. Ѱ Baynetworks 450T 24-way 100 Mbits/sec switch

        2. ΰ 12-way /Ű/콺 switches

        3. 4 500 kVa Uninterruptible Power Supplies from APC

        4. Ѱ , Ű, 콺  忡  54 GB
           UW-SCSI2 ϵ ũ

     COCOA ü ΰ?
        Linux! Ư   5.1 ߴ.
        <http://www.redhat.com> Ӱ   ִ
        ṵ̈  i386 ȣȯǴ ǻͿ ۵ 
        ִ.( PowerPC, Alpha, Sparc, Mips, Ataris Amigas
        ۵Ѵ.)  Ŀ Linus Torvalds torvalds@transmeta.com
        ٸ ڿڵ鿡   .  Ǵ 
        α׷ Ϲ н ̸ κ  GNU
        Ʈ .

     COCOA  Ʈ ġǾ?
           Ʈ ġǾ.

        1.  5.1 ϴ ⺻ ٷ̵
           <http://www.redhat.com>

        2. GNU C/C++ Ϸ Ƽ ȭ GNU C/C++ Ϸ
           (gcc, pgcc)

        3. Fortran 77/90 Ϸ Portland ׷ ϴ 

        4. C/C++, Fortran77/90 ó   ְ ִ
           Message Passing Interface (MPI) ̺귯

        5. Amtec Corporation  <http://www.amtec.com> ϴ
           TECPLOT Scientific Visulaization Ʈ

     COCOA ΰ?
        뷫 100,000̴.

  2.  COCOAۿ  󼼻

  2.1.  ϵ ġ

  ϵ ġϴ     ִ. ֿ  :

  1.  ӽŵ ڽ  ġ   ְ ȣ
     δ.

  2. 24-Ʈ Ʈũ ġ ġϰ 10Mbit Ʈ߿ ϳ 
     ι° ̴ Ϳ Ͽ 缳 Ʈũ . 
     23 Ʈ Ŭ̾Ʈ ̴ Ϳ Ѵ. 2 Ʈ
      ִ Ȯ ī带 ִ 2 Ŭ̾Ʈ  ġ
     ߰Ѵ.

  3. 16-way /Ű带 ġϰ ȭ κа 25 ӽ 
     Ű ̺ ϰ  Ѵ.

  4. 4 UPS Ŀ Ѵ.

  2.2.  Ʈ ġ

  ⼭   ۾ ϴ! ֿ:

  1.  켱 ġѴ.  5.1 CD-ROM ̿Ͽ ġѴ.
     κ ϵ ڵ Ǹ (Ʈũ ī带 Ͽ)
     ߿  ̺긦 ϰ ġ ʿ ٷ̸ Ѵ. 3
     GB root Ƽ ý   ġ Ű  
     д. ΰ 128 MB  Ƽ Ű  50 GB
     ڸ  ܵд. /tmp  1 GB Ҵϴ  
     ̶  ߿  ̴.

  2. ֽ  Ŀ(⼭ #2.0.36)   Ƽ GNU CC
     Ϸ pgcc  <http://www.goof.com/pcg/> ( Ƽ II ȭ
     ڵ带 Ŵ) ̿Ͽ SMP  ϵ ˸ û
     Ͽ Ŀ Ѵ.  ȭ û׵ ȴ:
     pgcc -mpentiumpro -06-fno-inline-functions SMP  ϴ 
     Ŀ  ޴   ִ Processor type and features ִ
     ư ν ϴ.( make xconfig ϸ鼭 ۵ȴ.

  3.   ο Ŀ- NFS  (knfsd)
     <http://www.csua.berkeley.edu/~gam3/knfsd/>  NFS 
       ġǾ - NFS  üϿ ġǾ
     ִ.  ġ ٶ 쿡  RPM ٷ̸
     <http://rufus.w3.org/linux/RPM/> Ͽ ġϸ ȴ. Ʈ
     û ȴ.

  4. ssh  <http://www.cs.hut.fi/ssh/>   ְ Ͽ
     ġ  ִ. (ܺη    ؼ) ssh-1.2.26
     ssh-2.0.11 ȣȴ. ֳϸ ssh v2.x  ȣȯ
     Ӵ   ̴. sshd  /etc/rc.d/rc3.d
     runlevel 3 Ѵ. ֱٿ ssh RPM
     <http://rufus.w3.org/linux/RPM/> ǰ ְ ٸ RPM 鵵
     ġϱⰡ  ִ.

  5. ΰ 3c905B ̴ ͵ Ǿ; Ѱ eth1μ IP
     128.118.170.11 Ǿְ ٸ ϳ 缳 Ʈũ
     Ǿ dummy IP ּҸ 10.0.0.1 ҴǾ ִ. Donald
     Becker   3COM 3c905B Ϳ  ֽ ̹
     (3c59x.c v0.99H
     <http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html>) Ŀγη
     ϵǾ 100Mbit/sec Full-duplex  Ѵ. ̴ vortex-
     diag ƿƼ ̿Ͽ
     <http://cesdis.gsfc.nasa.gov/linux/diag/vortex-diag.c> ̿Ͽ
     Ȯ  ִ.  ؼ   ؾѴ:
     /etc/sysconfig/network, /etc/sysconfig/network-scripts/ifcfg-eth0
     /etc/sysconfig/network-scripts/ifcfg-eth1. ⿡  ̴:

     /etc/sysconfig/network:

     NETWORKING=yes
     FORWARD_IPV4=no
     HOSTNAME=cocoa.ihpca.psu.edu
     DOMAINNAME=ihpca.psu.edu
     GATEWAY=128.118.170.1
     GATEWAYDEV=eth1
     NISDOMAIN=ihpca.psu.edu

     /etc/sysconfig/network-scripts/ifcfg-eth0:

     DEVICE=eth0
     IPADDR=10.0.0.1
     NETMASK=255.255.255.0
     NETWORK=10.0.0.0
     BROADCAST=10.0.0.255
     ONBOOT=yes
     BOOTPROTO=none
     USERCTL=no

     /etc/sysconfig/network-scripts/ifcfg-eth1:

     DEVICE=eth1
     IPADDR=128.118.170.11
     NETMASK=255.255.255.0
     NETWORK=128.118.170.0
     BROADCAST=128.118.170.255
     ONBOOT=yes
     BOOTPROTO=none
     USERCTL=no

  6.  ġ , BOOT  ̿ؼ Ʈũκ PC
     ϳϳ ϱ ߴ. BOOTP  /etc/inetd.conf
       ּ ν ϴ.

     bootps  dgram   udp     wait    root    /usr/sbin/tcpd  bootpd

    ÷Ǵ 3c905B Ʈũ ͸ ϴ Ŀ ̿ؼ
  غ  ְ  ʹ  Ŭ̾Ʈ   ǰ 
   96-bit Ʈũ ּұ ˷ش. (  00C04F6BC052) 
  ּҸ ̿ϸ /etc/bootptabϿ   ؾѴ:

          .default:\
                  :hd=/boot:bf=install.ks:\
                  :vm=auto:\
                  :dn=hpc.ihpca.psu.edu:\
                  :gw=10.0.0.1:\
                  :rp=/boot/client/root:

          node1:ht=ethernet:ha=00C04F6BC0B8:ip=10.0.0.2:tc=.default
          node2:ht=ethernet:ha=00C04F79AD76:ip=10.0.0.3:tc=.default
          node3:ht=ethernet:ha=00C04F79B5DC:ip=10.0.0.4:tc=.default
          .
          .
          .
          node25:ht=ethernet:ha=00C04F79B30E:ip=10.0.0.26:tc=.default

  7. /etc/hosts   Ѵ:

             127.0.0.1       localhost       localhost.localdomain
             # Server [COCOA]
             128.118.170.11 cocoa.ihpca.psu.edu cocoa.aero.psu.edu cocoa

             # IP address <--> NAME mappings for the individual nodes of the cluster
             10.0.0.1        node0.hpc.ihpca.psu.edu node0           # Server itself!
             10.0.0.2        node1.hpc.ihpca.psu.edu node1
             10.0.0.3        node2.hpc.ihpca.psu.edu node2
             .
             .
             .
             10.0.0.26       node25.hpc.ihpca.psu.edu node25

  /etc/hosts.confϿ    ϰ Ѵ:

          order hosts,bind

  ̴ DNS    û ֱ /etc/hostsϿ ִ IP
  ּҸ  ã ش.

  8.  Ͻý۵ /etc/exports   ߰ϸ ȴ:

             /boot           node*.hpc.ihpca.psu.edu (ro,link_absolute)
             /mnt/cdrom      node*.hpc.ihpca.psu.edu (ro,link_absolute)
             /usr/local      node*.hpc.ihpca.psu.edu (rw,no_all_squash,no_root_squash)
             /home1          node*.hpc.ihpca.psu.edu (rw,no_all_squash,no_root_squash)
             /home2          node*.hpc.ihpca.psu.edu (rw,no_all_squash,no_root_squash)
             /home3          node*.hpc.ihpca.psu.edu (rw,no_all_squash,no_root_squash)
             /home4          node*.hpc.ihpca.psu.edu (rw,no_all_squash,no_root_squash)

  9.  Ŭ̾Ʈ 忡   ġ ,  5.1
     KickStart ġ ̴̻.   kickstart 
     /boot/install.ks  :

             lang en
             network --bootproto bootp
             nfs --server 10.0.0.1 --dir /mnt/cdrom
             keyboard us
             zerombr yes
             clearpart --all
             part / --size 1600
             part /local --size 2048
             part /tmp --size 400 --grow
             part swap --size 127
             install
             mouse ps/2
             timezone --utc US/Eastern
             rootpw --iscrypted kQvti0Ysw4r1c
             lilo --append "mem=512M" --location mbr
             %packages
             @ Networked Workstation
             %post
             rpm -i ftp://10.0.0.1/pub/CLUSTER/RPMS/wget-1.5.0-2.i386.rpm
             rpm -i ftp://10.0.0.1/pub/CLUSTER/RPMS/xntp3-5.93-2.i386.rpm
             /usr/bin/wget ftp://10.0.0.1/pub/CLUSTER/kernel/vmlinuz -O/boot/vmlinuz
             /usr/bin/wget ftp://10.0.0.1/pub/CLUSTER/conf/lilo.conf -O/etc/lilo.conf
             /sbin/lilo
             /usr/bin/wget ftp://10.0.0.1/pub/CLUSTER/conf/hosts.equiv -O/etc/hosts.equiv
             sed "s/required\(.*securetty\)/optional\1/g" /etc/pam.d/rlogin > /tmp/rlogin
             mv /tmp/rlogin /etc/pam.d/rlogin

   KickStart ġ    
  <http://www.cache.ja.net/dev/kickstart/KickStart-HOWTO.html> 
   ֽϴ.  ġ߿ /etc/pam.d/rlogin ó
     Ѵ:

          auth       optional     /lib/security/pam_securetty.soauth       optional     /lib/security/pam_securetty.so

  ̰ rlogin/rsh ۽  Ŭ̾Ʈ н 
  ϰ Ѵ. ̴ Ŭ̾Ʈ  Ʈ   ̴.
   /etc/hosts.equiv  :

          node0
          node1
          node2
          node3
          .
          .
          .
          node25

    5.1 CD-ROM NFS  Ŭ̾Ʈ ϴ 
  /mnt/cdrom Ʈ Ű SMP ϴ ο Ŀ  
   ϰ  Ʈ Ͽ ִ  Ŀ ٲ۴. 
  Ŀ Ŭ̾Ʈ 忡 ϱ  Ư  ɼ ʿ
  ʴ.  BOOTP ̿ؼ Ŀ  ڵ  ɼ
  Ŀμ޴ Ʈŷ û׿ ϰ   ִ. ̴
  尡 ڵ ð߿   ip ο  䱸ȴ.
  Ʈ Ͽ   KickStart 忡 
  ϱؼ ȴ.  Ŭ̾Ʈ   ʿ  
  Ʈ ũ ְ ũ̼ǿ  Ű ڵ 
   ٸ ȴ.

  10.
      Ŭ̾Ʈ ġ õڸ, Ŭʹ ۵
     ϰԵȴ. brsh(
     <http://www.beowulf.org/software/RPMS/beobase-2.0-1.i386.rpm>)
       ƿƼ  Ŭ̾Ʈ 鿡 rsh ̿
       ϱ ġȴ. NIS  Ŭ̾Ʈ 忡
      α ϱ  ġ   ſ  
     ũƮ ̿ؼ  /etc/passwd, /etc/shadow
     /etc/group й  ִ.

  11.
     κ 񽺵  Ŭ̾Ʈ 鿡  /etc/inetd.conf
      Ұϰ ǰ  Ŭ̾Ʈ    ̰
     ȴ:

             shell   stream  tcp     nowait  root    /usr/sbin/tcpd  in.rshd
             auth    stream  tcp     nowait  nobody  /usr/sbin/in.identd in.identd -l -e -o

  12.
     automount ٷ̴  忡 ġϿ 䱸  پ 
     Ƽ ڵ Ʈ ش. ̴   NFS 
     , װ  Ҿϴ.   
     automount غ  Ǿ ʰ Ϲ NFS Ʈ 
     ŵȴ.

  13.
     MPI-CH ҽ ڵ  <http://www.mcs.anl.gov/mpi/>  
      pgcc ̿Ͽ ϸ ȴ. /usr/local/ Ƽǿ
     ġϴ  ū   ̴. mpif77ũƮ 
     䱸 ° Ǿ  mpif90  .
     /usr/local/mpi/util/machines/machines.LINUX  Ŭ̾Ʈ 
     ΰ Է ϸȴ. (dual-processor SMP 带 ϴ 쵵
     .) mpirun ̿Ͽ α׷ Ѵ.

  14.
     queing ý, DQS v3.0
     <http://www.scri.fsu.edu/~pasko/dqs.html> ٿε   ְ
     Ͽ /usr/local/DQS/ ġǰ NFS ̿Ͽ 
     Ŭ̾Ʈ ̿ؼ ٰ ϰ Ѵ.   Ŭ̾Ʈ
     ȭ /etc/services 񽺵 ÷ϰ  qmaster
     Ŭ̾Ʈ dqs_execd Ͽ ۵ϰ   ִ. 
       ĥ  ִ. ̴ DQS ߸ ȭ̴.
     ڰ DQS  Ͽ Ŭ̾Ʈ 尡 νϵ ϱ ʹ
      ð ɷ, ѹ       ִ.
     Wrapper  ũƮ DQS  ߸ ϰ  
     Ű  ʿ qsub, qstat qdel   ڰ Ǿ.
       qstat  Ͽ queue ִ  ۾ 
     ûǴ   ְ .  ٸ 
     qinfo,qload qmem   ؼ  load ͸ ְ
      Ŭ ҽ ̿뿡 ǰ .

  15.
     COCOA  ̰ ġŷ  ó  غ
     Ǿִ. Ŀ  pgccϷ ̿ϴ  C/C++ ڵ带
      ȴ. Ư  FPU intensive number crunching 带
      pgcc -mpentiumpro -06 -funroll-all-loopsɼǸ ̿ϸ
     gcc ̿ϴ   ð 30%  ´.

       Anirudh Modi <htmlurl url="mailto:anirudh-modi@psu.edu" name="anirudh-modi@psu.edu">  ȴ.  ְų Ȼ  ڿ  ֽʽÿ.

