OSLAF Setup

OSLAF Topology

[top]

OSLAF - Computer and Network Hardware

Single Master Node:

SuperMicro 370DLE Mother board (max. 1 GHz P3, 4 GB RAM)
Dual Pentium III, 733 MHz
512 MB RAM
IBM 20.5 GB IDE HD, CD-ROM drive, floppy drive
2 Intel 10/100 82559 Fast Ethernet cont. (1 onboard)
1 Apaptec SCSI PCI card
1 ATI XP98 Rage XL 4 MB Graphics Card
(no expansion slots left)

10 identical Slave Nodes:

Tyan Tiger 133 Motherboard (max. 1 GHz P3, 1.5 GB RAM)
Dual Pentium III, 733 MHz
512 MB RAM
IBM 20.5 GB IDE HD, CD-ROMM drive, floppy drive
1 Intel 10/100 82556 Fast Ethernet PCI card
1 ATI XP98 Rage XL 4 MB Graphics Card
(no expansion slots left)

Ethernet Switch:

3Com Superstack II, 24 ports, 100 Mbit

[top]

OSLAF - Other Hardware

KVM (Keyboard Video Mouse) Switch:

Belkin OmniView PRO 16 Port

2 Surge Protectors:

Belkin 10-Outlet Rack Mount Surge Protector
(all parts of OSLAF are powered via these, 25000 $ connected equipment warranty!)

To log on at the console:
— switch on monitor
— select channel 0 at the OmniView
— move mouse to wake up GUI

[top]

Special C3 (Cluster Command Control) Commands
The Cluster Command Control(C3) tools are a suite of cluster
tools developed at ORNL.  The suite included tools for cluster-
wide command execution, file distribution and gathering, process
termination, remote shutdown and restart, and system image updates.

Since the master node is NSF mounted on all Oslaf slave nodes, users
normally will only use the cexec command.  

All C3 commands have man pages.
cexec:  general utility that enables the execution of any standard command on
        cluster nodes. (Oslaf nodes are numbered 0 through 9).
        examples:  cexec ls -l     //execute ls -l on all nodes
                   cexec :0 ls -l  //execute ls -l on node 0 only
                   cexec :1-4,6 ls -l //execute ls -l on nodes 1 thru 4 and node 6.


cget:  retrieves files or directories from all or selected nodes


ckill: terminates a user specified process on cluster nodes


cpush:  distribute files on cluster nodes


cpushimage: update system image on nodes using image from SystemImager tool.


crm: remove files or directories from all cluster nodes


cshutdown: shutdown or restart cluster nodes.


cnum: returns a node range number based on node name.


cname: returns node names based on node ranges.

clist: returns all clustera nd their type in a configuration file.
 

[top]

Monitoring System Status and Performance

There is a nice browser monitor using netscape to oslaf.physics from,
e.g. doco. You'll need to get the account name and password
More about this later.

[top]



Back to OSLAF menu

 

Revision date: 07-Jul-2003