SPCM (Simple, Portable Cluster Manager is a free), open source integrated tool set for managing a simple HPC (High Performance Computing) cluster.
It is the only portable cluster management suite we are aware of and is designed to be easily adapted to most POSIX platforms.
SPCM automates the process of configuring a head node, compute nodes, file servers, and visualization nodes. Most common management tasks can be performed using a simple menu interface, while additional tasks are supported by command-line tools.
SPCM automatically installs and integrates the LPJS scheduler and the Apache web server, ready to accommodate the monitoring system of your choice (Nagios, Munin, Prometheus, etc.)
Developed by Jason W. Bacon and James F. Wagner
- Simple menu interface for most common operations
- Fast, easy installation and node deployment
- Designed for portability to any POSIX platform
- Remote power control using IPMI
The design philosophy centers on simplicity and performance. These ideals are achieved in-part by minimizing interdependence of cluster nodes. Each compute node contains a fully independent operating system installation and critical software installations on its own local storage. Compared with clusters that utilize shared storage more extensively, this strategy increases cluster setup and maintenance time slightly in exchange for simpler management, less "noise" on the local network, fewer single points of failure, and fewer bottlenecks.
-
Simplicity: Efforts are focused on objective measures such as reliability, speed, easy management. No peacock feathers that would drain man-hours from improving functionality and add needless complexity and more bugs. We intend to keep the "Simple" in SPCM and will not let it fall victim to creeping feature syndrome.
-
Portability: SPCM should be easy to port to any POSIX operating system. This is a challenge given the differences in sysadmin tools across various POSIX systems, but the basic design minimizes barriers to supporting different tools. The long-term plan is to support heterogeneous clusters, where different nodes can opaquely run different operating systems, while being managed with the same tools. There is currently limited support for this and using FreeBSD file servers and visualization nodes
-
Non-interference with core operating system: Unlike many cluster management systems, SPCM does not depend on hacks to the base operating system, but sits on top of the standard base system and package manager for each OS. Critical security updates just released for your OS? Install them immediately without fear of breaking your cluster: You don't have to wait for us to update a custom OS image, leaving your cluster vulnerable in the meantime.
-
SPCM clusters never need to be shut down: All system updates can be applied to a live cluster. Compute nodes are set to draining state and updated when they become idle. The head node, file servers, and visualization nodes can all be updated and rebooted via the menu interface without adversely affecting running jobs.
Implementation of this design is facilitated by leveraging the systems management tools provided by the base system, including native package managers and cross-platform package managers.
The SPCM tools are written almost entirely in POSIX Bourne shell using standard Unix tools to configure the system, and utilizing ports/packages for all software management.
From a networking perspective, a typical HPC cluster is a LAN (Local Area Network): A group of computers on a private network behind a router configured with NAT (Network Address Translation). The difference between your home or office LAN and an HPC cluster is in the power of the nodes, the speed
In theory, an HPC cluster could simply be a subset of nodes on a general-purpose network. However, using a dedicated switch behind a router isolates all the cluster network traffic, which will improve performance for file server access and parallel program communication, while also eliminating impact on the outside network.
In many clusters, the head node is multi-homed (has two network interfaces) and serves as the gateway for the cluster. SPCM allows for this configuration, but be aware that it complicates the setup of the head node as well as configuration of many services running
The recommended hardware configuration uses a single network interface on every node, including the head node, a separate router/gateway, and a dedicated switch for all cluster traffic. Many modern network switches have built-in routing capability and can serve as both router and local switch. If you're using a simple switch without routing capability for your cluster, you can use an inexpensive hardware router or quickly and cheaply build a sophisticated firewall router using any PC with two network adapters and OPNsense.
A dedicated router appliance is also easier and likely more secure than a head node configured as a gateway. Tools like pfSense are written and maintained by networking experts and provide a convenient web interface for configuration.
In addition, this topology allows direct connection from outside the cluster to any node via port forwarding with different TCP ports. No need to run additional cables to isolate large transfers to/from the file servers. For example, incoming SSH connections on port 22 can be routed to the head node, while connections on port 22001 can route to a file server. In this way, the head node is spared the network load of file transfers, which can annoy interactive users. On a large cluster, you can use faster network interfaces on both the WAN and LAN side of the router to support the full bandwidth of multiple servers within the cluster. E.g., if your cluster nodes all use gigabit interfaces, using 10 GbE on the router will support multiple file transfers to/from different servers at full gigabit speed while still leaving plenty of bandwidth for the head node.
If you plan to use the SPCM PXE installer for non head nodes, you'll probably want to disable the DHCP server on the router. SPCM can automatically configure your head node as a DHCP and PXE server.
It's best to keep the load on the head node as low as possible to ensure snappy response times for shell sessions and for the scheduler. Hence, the head node should not double as a compute node or as a file server for large amounts of data. We generally house /home on the head node so that it is fully functional even when all other nodes are down, but with a very small quota (e.g. 250 MiB). Scientific data are stored on separate file servers so that heavy network traffic and disk loads are isolated.
A compute node can double as a file server and this has the advantage that jobs running on that node have direct access to the disks rather than using the network via NFS, Gluster, etc. If you do this, be aware that ZFS by default will consume most or all available RAM, thus competing with computational processes. To prevent performance problems, you can limit ZFS adaptive read cache (ARC) to a few gigabytes and subtract the same amount from RealMemory for that node in your slurm.conf.
Accessing the head node and file servers via NFS may mean cross-mounting them (each is an NFS server for the other), which can cause a deadlock during boot while each waits for the other to enable NFS. On FreeBSD, this issue is easily solved using background mounting (bg flag in /etc/fstab). Note, however, that background mounting does not work on RHEL/CentOS 7 due to incompatibilities with systemd, so if you cross-mount, you will need a more complex setup using autofs or the noauto mount flag + a cron job for late mounting.
Visualization nodes should be considered a tool for quick-and-dirty viewing of results. For more sophisticated viewing, users should download the data to a workstation where they can utilize the local display for best graphics performance.
The head node and file servers should generally be on battery back-up, but not the compute nodes. Keeping compute nodes running through a power outage would require a truckload of batteries for a large cluster and greatly reduce battery run-time for even a very small cluster.
The head node need not be powerful, but should be very reliable. For large clusters, it is recommended that the head node have redundant power supplies and boot from a RAID with hot swappable disks. ZFS can be utilized to construct a RAID without the need for a hardware RAID controller, though replacing disks in a software RAID is a little more involved. SLURM may use a fair amount of RAM on the head node of a large, busy cluster, but does not need much CPU.
For a small personal cluster, a laptop actually makes a pretty good head node with its built-in battery backup, keyboard, and monitor.
File servers should be similarly reliable, with redundant power supplies and RAID. You may want to equip them with faster network interfaces than the compute nodes, as they may be greatly outnumbered, or a gigabit network interface may be a bottleneck next to the RAID capability. Our benchmarks showed little difference in performance between SAS and SATA disks. If you use SATA, however, be sure that they are server-grade. Low-end SATA disks designed for PCs may not offer the same performance or may not be rated for use in large RAIDs due to vibrational characteristics. File servers should have plenty of RAM for buffering to allow reordering of I/O operations. A few fast processors will serve better than many slower ones in most settings.
Don't obsess about hardware specs. Predicting the optimal CPU and memory configuration for the variety of programs that will run on the cluster is impossibly complex. Be content that you are reducing weeks, months, or years of computation time to hours or days in any case, and go with what looks like the most cost-effective CPUs and RAM. Doubling the cost of a cluster for the 20% gain in speed you get from the latest-and-greatest processors is usually nothing but foolish ego fodder.
Do choose an expandable configuration, e.g. don't fill all of your memory slots right away, in case you find out more RAM is needed later. Four 16 GiB chips may cost a little more than eight 8 GiB chips, but filling all your DIMM slots with 8 GiB chips will backfire later if you have to replace them all in order to increase total RAM.
Maximizing rack density is also generally a good idea for multiple reasons. Compute nodes with a higher core count save capital costs on both the servers and rack space, save energy, and reduce network load since processes running on the same node can communicate through local memory rather than the network.
Resources for compute nodes should be put toward CPU and RAM. Redundant power supplies are of little use unless interruption of jobs would be catastrophic in your setting. Maximizing MIPS/$ is usually a good strategy for general-use clusters, so the highest-end CPUs are usually not a good value. 16 medium-speed cores will generally be a better value than 8 high-speed cores, unless you need to run many jobs that don't scale well to large numbers of processes.
Redhat Enterprise Linux (RHEL) and it's derivatives are the de facto standard operating systems for HPC clusters. They are more stable than bleeding-edge Linux distributions, have strong support for HPC system software like Infiniband drivers, parallel file systems, etc., and are the only POSIX platforms officially supported by most commercial scientific software vendors.
The main disadvantages of enterprise Linux platforms (compared to FreeBSD or community Linux distributions such as Debian and Gentoo) are use of outdated kernels and packages available in the Yum repository. (Stability and long-term binary compatibility in enterprise Linux systems is maintained by running older, time-tested, and heavily patched versions of system software.)
FreeBSD's unparalleled reliability, near-optimal efficiency, and easy software management via the FreeBSD ports collection make it an ideal platform for HPC clusters. There is no better platform for running scientific software that requires modern development tools or lengthy uninterrupted up time. FreeBSD is the only operating system we've found that offers enterprise reliability and system management features (binary updates, fully-integrated ZFS, etc) combined with top-tier development tools and software management (Clang/LLVM base compiler, FreeBSD ports, etc.).
An example of FreeBSD's reliability is provided by Peregrine, a FreeBSD HPC cluster built for educational use at the University of Wisconsin -- Milwaukee. Peregrine has never had a node crash or freeze in the absence of a hardware problem in 8 years of service, despite running some extremely intensive jobs that caused outages on other clusters. The only reliability issues encountered were a few head node crashes, traced to a Dell PowerEdge firmware bug affecting single-processor systems, and a compute node crash caused by a bad memory slot.
FreeBSD is the basis of many products used in HPC including FreeNAS, Isilon, NetApp, OPNSense, Panasas, and pfSense.
Many FreeBSD HPC clusters are in use today, serving science, engineering, and other disciplines. FreeBSD is a supported platform on Amazon's EC2 virtual machine service. It is also a little-known fact that the special effects for the movie "Matrix" were rendered on a FreeBSD cluster.
FreeBSD can run most Linux binaries natively (with better performance than Linux in some cases), using its CentOS-based Linux compatibility module. This module is NOT an emulation layer. It simply adds Linux system calls to the FreeBSD kernel so that it can run Linux binaries directly. Hence, there is no performance penalty. The only added cost is a small kernel module and modest amount of disk used to house the module and Linux software.
- Unix systems management and security (ssh setup, filesystem management, user management, software installs, etc)
- Basic networking (host configuration, router setup, LAN, port forwarding, DHCP server)
- NFS server and client configuration
- Basic web server configuration
Do a basic RHEL minimal or FreeBSD installation on your head node. Configure the network as desired. Either place all nodes behind a separate gateway (recommended) or configure the head node as a gateway (somewhat more involved, see Design tab). A LAN address of 192.168.x.2 is recommended for the head node. Be sure to choose a LAN IP range that does not overlap with the WAN.
On FreeBSD, ZFS is recommended for the head node unless is has less than 4GB RAM, and for all file servers. UFS2 is recommended for compute nodes since it requires far less RAM than ZFS, leaving more for your scientific computations.
Download and run the spcm-bootstrap script from this repository to begin head node setup.
Once the head node is bootstrapped, run cluster-node-admin and follow the menu options for installing and configuring nodes beginning with the head node.
File servers should generally be done before compute nodes if possible, so that the compute node NFS mounts can be configured automatically during setup.
Run auto-pxe-node-discover ("PXE install new nodes" from the Node management menu of SPCM) BEFORE booting new nodes for PXE installation, and complete the DHCP configuration of each new node before it reboots following install (usually by simply accepting the defaults offered by auto-pxe-node-discover). This will permanently reserve the first IP address granted to node MAC in dhcp.conf.
Otherwise, the DHCP server may create duplicate leases for the same MAC address on subsequent boots. This will require stopping the DHCP server and manually cleaning up the MAC/IP links in dhcp.leases, dhcp.conf, and /etc/hosts. The cluster-dhcp-cleanup script can be used to do this in an orderly fashion.
Any nodes not installed by the PXE system will have to be manually configured to accept password-less ssh from the head node. Set PermitRootLogin=prohibit-password in sshd_config to allow this without compromising security.
The default PXE installer script is designed to be minimalist so you can quickly get multiple nodes up and running (and spend minimal time in the data center). New nodes are accessible from the head node, but have very little software installed. Subsequent steps can be performed remotely from the comfort of your office.
The "Initialize new nodes" option installs basic system software, updates the OS, and reboots. The menu system allows you to select a batch of any number of nodes to initialize unattended.
The final step, "Synchronize nodes", is done after initializing and rebooting (and possibly after a node has been offline for an extended period and needs to catch up to the other nodes on software installs and updates).
One node of each type (file server, visualization, compute) must be manually configured in this step to set up NFS mounts, etc. Subsequent nodes can be synchronized unattended using the configuration of the first.
While synchronizing one set of nodes (e.g. 10 or 20), you can run the initialization step on the next set. By overlapping configuration this way, you can reduce the waiting time for configuring all nodes. You might ask "why not sync multiple nodes in parallel?". The main reason is that doing so would overwhelm the network while downloading packages and updates.