Revision as of 23:54, 22 February 2011

Design Typologi

Cluster typologi

Headnode

Installation af SSH nøgle

Gør som ved her.

Ændre source.list til local mirror

Ændre source.list til den defineret her.

aptitude update
aptitude dist-upgrade

Installation af DHCP Server

aptitude install dhcp3-server -y

tilføj 192.168.149.0 scopet i config filen: /etc/dhcp3/dhcpd.conf

subnet 192.168.149.0 netmask 255.255.255.0 {
  range 192.168.149.50 192.168.149.200;
  option domain-name-servers 172.16.4.66;
  option domain-name "cluster.tekkom.dk";
  option routers 192.168.149.1;
  default-lease-time 600;
  max-lease-time 7200;
  filename "pxelinux.0";
  next-server 172.16.4.83;
  host Node1 {
    hardware ethernet 00:0b:db:92:d5:f9;
    fixed-address 192.168.149.101;
  }
  host Node2 {
    hardware ethernet 00:0B:DB:92:D5:3C;
    fixed-address 192.168.149.102;
  }
  host Node3 {
    hardware ethernet 00:0B:DB:92:D5:1E;
    fixed-address 192.168.149.103;
  }
  host Node4 {
    hardware ethernet 00:0F:1F:65:EC:B1;
    fixed-address 192.168.149.104;
  }
  host Node5 {
    hardware ethernet 00:0B:DB:92:D5:E4;
    fixed-address 192.168.149.105;
  }
 host Node6 {
    hardware ethernet 00:0B:DB:92:D5:E2;
    fixed-address 192.168.149.106;
  }
  host Node7 {
    hardware ethernet 00:0B:DB:92:D5:50;
    fixed-address 192.168.149.107;
  }
  host Node8 {
    hardware ethernet 00:0B:DB:92:D6:72;
    fixed-address 192.168.149.108;
  }
  host Node9 {
    hardware ethernet 00:0B:DB:92:D5:A0;
    fixed-address 192.168.149.109;
  }
  host Node10 {
    hardware ethernet 00:0B:DB:92:D5:E0;
    fixed-address 192.168.149.110;
  }
}

Installation af apt-cacher

aptitude install apt-cacher -y

har droppet det og bruger den vi har i forvejen

Installation af OpenMPI

aptitude install -y libopenmpi-dev openmpi-bin openmpi-doc

Installation af NFS server

Installerer en nfs server og laver et nfs export i /var/mirror samt brugernes home folder.

aptitude install -y nfs-kernel-server
mkdir /var/mirror

Tilføj folderen i /etc/exports

/var/mirror     *(rw,sync,no_subtree_check)
/home     *(rw,sync,no_subtree_check)

Genstart NFS Serveren

/etc/init.d/nfs-kernel-server restart

Installation af Distributed Shell

Nogle af de script til administration bruger dsh, så derfor skal det også installeres

aptitude install -y dsh

Installation af geany

aptitude install geany xterm -y

Installation af Nodes

Man kan installere nodes med nedenstående script. Man kalder det bare med ip og navn på maskinen som henholdvis 1. og 2. argument.

Installations scripts

#!/bin/bash
# Laver en .ssh folder og kopierer nøgler ind i den på node
ssh $1 mkdir .ssh
scp .ssh/* $1:.ssh/

#Tilføjer ip og navn til hosts filen, Client filen og mpd filen.
echo $1 $2 >> /etc/hosts
echo $2 >> Clients.txt
echo $2 >> mpd.hosts

#kopierer hosts filen ud til alle clienter, så de kan svare tilbage til ClusterH
scp /etc/hosts $2:/etc/hosts

#Sætter clienterne på til at bruge det lokale mirror
scp /etc/apt/sources.list $2:/etc/apt/sources.list

#Omdøber klienten
ssh $2 "echo $2 > /etc/hostname"

#Initializerer med den nye mirror server, og installerer nødvendige SW
ssh $2 "aptitude update && aptitude install -y openmpi-dev openmpi-common nfs-common openmpi-bin"

#Sætter NFS Rettigheder
scp /etc/default/nfs-common $2:/etc/default/nfs-common
ssh $2 "/etc/init.d/idmapd restart"
ssh $2 "/etc/init.d/rpc_pipefs restart"

#Opretter og mounter /var/mirror og /home
scp ./mount.txt $2:./
ssh $2 "mkdir /var/mirror"
ssh $2 "cat ./mount.txt >> /etc/fstab"
ssh $2 "mount -a && reboot"

mount.txt filen

ClusterH:/var/mirror/ /var/mirror nfs rw 0 0
ClusterH:/home/ /home nfs rw 0 0

nfs-common filen

Har ændret NEED_IDMAPD til "yes"

og så skal /etc/init.d/idmapd & /etc/init.d/rpc_pipefs genstartes efter.

# If you do not set values for the NEED_ options, they will be attempted
# autodetected; this should be sufficient for most people. Valid alternatives
# for the NEED_ options are "yes" and "no".

# Do you want to start the statd daemon? It is not needed for NFSv4.
NEED_STATD=

# Options for rpc.statd.
#   Should rpc.statd listen on a specific port? This is especially useful
#   when you have a port-based firewall. To use a fixed port, set this
#   this variable to a statd argument like: "--port 4000 --outgoing-port 4001".
#   For more information, see rpc.statd(8) or http://wiki.debian.org/?SecuringNFS
STATDOPTS=

# Do you want to start the idmapd daemon? It is only needed for NFSv4.
NEED_IDMAPD=yes

# Do you want to start the gssd daemon? It is required for Kerberos mounts.
NEED_GSSD=

NTP opsætning

Jeg har sat headnode op til at opdatere tiden på alle nodes. Uden nogen god grund:-)

Ændre lidt i crontab med crontab -e

# m h  dom mon dow   command
*/10 * * * * ~/RunOnClients.sh "ntpdate 192.168.139.29"
*/10 * * * * /usr/sbin/ntpdate 192.168.139.29