IGGI Project (doc: 22/11/2006, ISO: 17/11/2006)

Antoine Ginies (aginies _at_ mandriva.com)

Revision History
Revision 2.4 21 Nov 2006 ag
add blcr chapter, imrpove PDF output
Revision 2.0 28 Nov 2005 ag
first major update
Revision 1.0 December 2002  
CLIC version. Original authors: Antoine Ginies, Erwan Velu (erwan.velu -_at-_ free point_ fr), Goneri LeBouder, Translators: Fabian Mandelbaum, Christian Roy, Robert Kulagowski, Sunil Patel
Revision 0.9 2 August 2002  
First public version, authors: Antoine Ginies, Erwan Velu

Table of Contents

1. Introduction, and where get ISO
1.1. What is the purpose of this document ?
1.2. Download IGGI ISO
1.3. What is a IGGI Cluster ?
1.4. IGGI partners
1.5. Some software inculded in IGGI release
1.6. Typical Server Architecture (CLIC mode)
1.7. Full list of RPMS available
1.8. Typical Node Architecture with dedicated HDD (CLIC mode)
1.9. Typical diskless Node
1.10. Installation procedure
2. HOWTO to add the cluster bundle into the CS4
2.1. CS4 + cluster repositery
2.1.1. Setting up the CS4+cluster bundle repositery
2.1.2. HTTP and NFS configuration
2.2. Prepare the CS4 server
3. Step by Step Installation
3.1. Mandriva Linux Installation Wizard
3.1.1. Selecting your language
3.1.2. Security Level
3.1.3. Partitioning
3.1.4. Packages Selection
3.1.5. Root Password
3.1.6. Network configuration
3.1.7. End of the Server Installation
3.2. IGGI Server Server Post-Installation Configuration
3.2.1. Copying and Preparing Media
3.2.2. Install directory
3.2.3. Server configuration
3.2.4. Configuring DNS
3.2.5. Configuring NIS
3.2.6. Configuring DHCP
3.2.7. Configuring PXE
3.2.8. Setup administration
3.2.9. Compute
3.2.10. Configuring Postfix (disable by default)
3.2.11. OAR database
3.2.12. End of the Automated Configuration
3.2.13. drakcluster to register MAC address
3.3. Collecting MAC Addresses (not mandatory but usefull)
4. Setting Up a DNS Server
4.1. /var/lib/etc/named.conf
4.2. /var/lib/named/var/named/zone Files
4.2.1. 127.0.0
4.2.2. db.12.12.12..hosts
4.2.3. db.guibland.com.hosts
5. Set Up a DHCP Server
6. Setup xinetd
6.1. /etc/xinetd.d/ files
6.1.1. tftp config
6.1.2. rsh config
6.1.3. distcc config
7. Setup NIS
7.1. NIS server with autofs map
7.2. Nis administration
8. PXE quick overview
8.1. PXE Documentation
9. Drakcluster User guide
9.1. Drakcluster user guide
9.2. Nodes page
9.3. Execute a command
9.4. Open terminal
9.5. Admin/compute
9.6. System log page
9.7. PXE configuration page
9.8. cluster_applet
10. installation and configuration of IGGI Node
10.1. Installing and Configuring the First Node
10.1.1. Which Boot Method to Use?
10.1.1.1. Network installation
10.1.1.2. CDROM
10.2. Installing the Node
10.2.1. Install process
10.2.2. Partitioning
10.2.3. Package Selection
10.2.4. Network Configuration
10.2.5. Setting a root Password
10.2.6. End of Installation
11. Diskless node
11.1. Create diskless image
11.2. use Dolly to copy the image in nodes RAM
12. Duplicate an operating System
12.1. Three way to duplicate a computer over the network
12.1.1. KA method
12.1.2. Dolly method
12.1.3. Dolly+ method
12.2. HOW it works
12.2.1. 3 steps
12.2.2. Needed files
12.3. Step 1: PXE, TFTP, DHCPD services
12.3.1. PXE parameters on server
12.3.2. Configuration files
12.4. Various scripts
12.4.1. clone_script
12.4.2. ka-d-session.sh script
12.5. Step 2: parrallel methods to get stage2
12.5.1. Get stage2 via KA method
12.5.2. Get stage2 via Dolly
12.5.2.1. Create a stage2.img
12.5.3. dolly.cfg configuration file
12.6. Step 3, the duplication process
12.6.1. Duplicate a golden node with KA
12.6.1.1. Know BUGS with desc file
12.6.2. Duplicate a computer with dolly
12.6.3. Duplicate a computer with dolly+
13. Administration of IGGI Nodes
13.1. setup_admin.pl script
13.2. Using Berkeley Tools
13.2.1. authd
13.2.2. gexec
13.3. ka Tools
13.3.1. ka-run
13.3.1.1. rshp
13.3.1.2. mput
13.3.2. Taktuk2
13.4. dssh
13.5. Tentakel
13.6. gsh
13.7. pssh
13.8. fanout
13.9. sauvegarde (Saving Data)
13.10. adduserNis.pl and deluserNis.pl
14. URPMI rpm downloader, installer and dependency solver
14.1. URPMI
14.1.1. What is URPMI?
14.1.2. What You Get With It?
14.2. Parallel Feature
14.2.1. Installing the Parallel Feature
14.2.2. Configuring the Nodes
14.2.3. The /etc/urpmi/parallel.cfg File
14.2.4. Running urpmi in Parallel
14.3. urpmi recover
15. OAR User guide
15.1. OAR user guide
16. Compute mode
16.1. Compute mode
17. Ganglia Monitor
17.1. Tools List
17.1.1. Definition
17.1.1.1. gmond
17.1.1.2. gexec
17.1.1.3. gstat
17.1.1.4. gmetric
17.1.1.5. gmetad
17.1.1.6. libganglia3-devel
17.2. Ganglia Installation
17.2.1. On a IGGI node
17.2.2. On the IGGI server
18. MPI Implementations
18.1. MPI
18.1.1. What is MPI?
18.1.2. The main implementations
18.2. MPICH-A Portable Implementation of MPI
18.2.1. What is MPICH ?
18.2.2. Current Release
18.2.3. How does it works ?
18.2.4. More documentation
18.3. LAM (Local Area Multicomputer)
18.3.1. What is LAM ?
18.3.2. Current Release
18.3.3. How does it works ?
18.3.4. More documentation
19. Other Parallel Implementations
19.1. PVM (Parallel Virtual Machine)
19.1.1. What is PVM?
19.1.2. Current Release
19.1.3. How Does it Work?
19.1.4. Test pvm
19.1.5. More Documentation
20. BLCR
20.1. Quick BLCR user guide
21. Quick testing
21.1. Testing the Cluster
21.1.1. Ganglia
21.1.2. Remote command
21.1.3. Simple MPICH Testing
21.1.4. Simple LAM Testing
22. Benchmark
22.1. Povray
22.2. Povray Mpi
22.3. How to benchmark?
22.4. Step by step example
22.5. Blender rendering farm
22.5.1. Server side
22.5.2. Nodes side
22.5.3. It's time to render !
23. GNBD, CLVM, GFS howto
23.1. Overview
23.2. Server side
23.3. Node side
23.3.1. Automatic mode
23.3.2. Join the fence domain
23.3.3. CLVM and gnbd services
23.4. CCS status
23.5. GNBD to export devices
23.6. Use the GNBD (Server on node side)
23.7. Create the Storage
23.8. GFS filesystem
23.9. Quick way to disable GNBD+GFS+CLVM

List of Figures

3.1. Select your language
3.2. By default, you should choose a "Standard" security level
3.3. Partitionning"
3.4. A Partitioning Example
3.5. A Server Package Selection
3.6. Individual Packages Selection
3.7. Choose a Root Password
3.8. A Sample Server Network Configuration
3.9. Setting the Hostname
3.10. Graphic interface not configured
3.11. Drakcluster initial configuration
3.12. Popup menu
3.13. How many MAC address
3.14. Wait for MAC address
3.15. MAC registered
3.16. Result in drakcluster
9.1. replicate a node
9.2. how many nodes to install
9.3. waiting for 2 nodes
9.4. Nodes's MAC registered.png
9.5. Start of KA
9.6. ka running
9.7. End of KA replication
9.8. Drakcluster state after adding to nodes
9.9. Execute a command on nodes
9.10. Open terminal
9.11. compute/admin
9.12. Drakcluster log page
9.13. Drakcluster PXE page
9.14. cluster_applet
10.1. Drakcluster PXE configuration
10.2. Example Node Packages Selection
10.3. Set the IP Attribution on DHCP
10.4. Choose a root Password
10.5. Congratulation Screen
12.1. KA stage1
12.2. Dolly stage1
21.1. Ganglia web page monitor
22.1. Blender rendering path
22.2. drqman
22.3. New blender job
22.4. Running job
22.5. Drqman computer