The requirements for the OpenQRM server are different from the requirements for the managed nodes. Following are the requirements for each:

Requirements for the openQRM server

Hardware 

The openQRM server runs on a uniprocessor or SMP machine with the following minimum characteristics:

• IA-32 Intel(R) Architecture (Pentium 4) or AMD OpteronTM or Xeon (EMT64)
  
• 1 GigaHertz processor
• 512 Megabytes RAM
• 500 Megabytes hard disk space Storage

The OpenQRM server must have access to storage capacity on a network file system (NFS), a network-attached storage (NAS) device over NFS, or storage area network (SAN) over NFS.

NFS storage can be local to the QRM server. However this is not recommended for production systems.

Note:
Required storage capacity depends on the space required by the boot and filesystem images that will be created. Approximate size of 3-4G per image.

Software

The software environment for the openQRM server software is tested on Red Hat Enterprise Linux Advanced Server 3 and Fedora Core 4. Additional distributions are currently being tested.

Networking

Stand-alone installation of OpenQRM requires one or two separate 100 base-T Ethernet (or better) network interfaces, one for internal use and one for external use for the network management of the organization.

• openQRM interface: Handles communications with the managed nodes in the cluster.
• LAN interface: Handles communications between the OpenQRM server and the outside network.

The openQRM server must have network access to all nodes under its management. 

Note that while openQRM can work quite well without the LAN interface, it is highly recommended that the installation is done with it.

Shared Storage Access 

All nodes must have access to the shared storage 

Requirements for the managed nodes

openQRM works by combining the resources of multiple physical machines. Each individual machine is known as a node. Each node must meet some basic requirements to participate in an openQRM network.

Following are the highlights of managed node boot process :
- Load the PXE-stack from the NIC
- Request an ip-address from a dhcpd-server (-> dhcpd-plugin)
- Download its PXE-config file from a tftp-server according its mac address (-> tftpd-plugin)
- Download kernel + initrd  as per PXE-config
- Then the node loads + starts the kernel, loads the intird, mounts its root-filesystem and does a pivot_root into it which init's the system.

Hardware Requirements

Each node managed by openQRM must be a uniprocessor or SMP server with the following minimum characteristics:

• IA-32 Intel(R) Architecture (Pentium 4) or AMD Opteron or Xeon (EMT64)
  
• 1 GigaHertz processor
• 512 Megabytes RAM
• Network interface Card(NIC) - Each node must have atleast one network interface. Nodes with two NICs is recommended.
  
• QRM interface(eth0): PXE boot compatible network interface card. This card needs to communicate with DHCP and TFTP server and also needs to respond to healthcheck messages from openQRM server.
• LAN interface(eth1): Standard network interface card.
  

Notes:

• The node machine's BIOS must be configured to boot from the PXE card.
• A node can operate without the LAN interface, but this is not recommended on production systems.
• Uses are assigned to particular interfaces here by way of example only. You can assign either connection to any interface.

Applications

Applications running on nodes managed by QRM must be able to run on the above available hardware under openQRM. Currently RedHat 7.3, 8.0. 9.0, RHEL 3 and 4 (both AS and ES), Fedora Core 4 as well as Suse 9 & 10 are supported. Additional distributions are currently being tested.

Comments (0)

Subscribe to this comment's feed