This lesson is still being designed and assembled (Pre-Alpha version)

Working on a remote HPC system

Overview

Teaching: 25 min
Exercises: 10 min
Questions
  • What is an HPC system?

  • How does an HPC system work?

  • How do I log on to a remote HPC system?

Objectives
  • Connect to a remote HPC system.

  • Understand the general HPC system architecture.

What is an HPC system?

The words “cloud”, “cluster”, and “high-performance computing” are used a lot in different contexts and with varying degrees of correctness. So what do they mean exactly? And more importantly, how do we use them for our work?

The cloud is a generic term commonly used to refer to remote computing resources of any kind – that is, any computers that you use but are not right in front of you. Cloud can refer to machines serving websites, providing shared storage, providing webservices (such as e-mail or social media platforms), as well as more traditional “compute” resources. An HPC system on the other hand, is a term used to describe a network of computers. The computers in a cluster typically share a common purpose, and are used to accomplish tasks that might otherwise be too big for any one computer.

/hpc-intro/Connect%20to%20cluster

Logging in

Go ahead and log in to the cluster: Saga at Uninett Sigma2 AS.

[user@laptop ~]$ ssh yourUsername@saga.sigma2.no

Remember to replace yourUsername with the username supplied by the instructors. You will be asked for your password. But watch out, the characters you type are not displayed on the screen.

You are logging in using a program known as the secure shell or ssh. This establishes a temporary encrypted connection between your laptop and saga.sigma2.no. The word before the @ symbol, e.g. yourUsername here, is the user account name that Lola has access permissions for on the cluster.

Where do I get this ssh from ?

On Linux and/or macOS, the ssh command line utility is almost always pre-installed. Open a terminal and type ssh --help to check if that is the case.

At the time of writing, the openssh support on Microsoft is still very recent. Alternatives to this are putty, bitvise SSH, mRemoteNG or MobaXterm. Download it, install it and open the GUI. The GUI asks for your user name and the destination address or IP of the computer you want to connect to. Once provided, you will be queried for your password just like in the example above.

The authenticity of host

When login in for the first time you may get a question whether you trust the server you are trying to connect. If you typed the address correctly (i.e. ) then it is safe to say “yes” to the question at the end of this message and permanently added this server to trusted hosts

$ ssh lola@
The authenticity of host ' (###.###.###.##)' can't be established.
RSA key fingerprint is SHA256:NwV2/9HMlLfj6hFmXTuA4UVievE/uq36K9EYa20CteI.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '' (RSA) to the list of known hosts.

Where are we?

Very often, many users are tempted to think of a high-performance computing installation as one giant, magical machine. Sometimes, people will assume that the computer they’ve logged onto is the entire computing cluster. So what’s really happening? What computer have we logged on to? The name of the current computer we are logged onto can be checked with the hostname command. (You may also notice that the current hostname is also part of our prompt!)

[yourUsername@login-2.SAGA ~]$ hostname
login-2.saga.sigma2.no

Nodes

Individual computers that compose a cluster are typically called nodes (although you will also hear people call them servers, computers and machines). On a cluster, there are different types of nodes for different types of tasks. The node where you are right now is called the head node, login node or submit node. A login node serves as an access point to the cluster. As a gateway, it is well suited for uploading and downloading files, setting up software, and running quick tests. It should never be used for doing actual work.

The real work on a cluster gets done by the worker (or compute) nodes. Worker nodes come in many shapes and sizes, but generally are dedicated to long or hard tasks that require a lot of computational resources.

All interaction with the worker nodes is handled by a specialized piece of software called a scheduler (the scheduler used in this lesson is called SLURM). We’ll learn more about how to use the scheduler to submit jobs next, but for now, it can also tell us more information about the worker nodes.

/hpc-intro/Connect%20to%20cluster

When a user login to the cluster using ssh, he or she landed on a login node, which act as an entrance point to the cluster. Login nodes are used for light weight tasks such as managing user login, copy files to the cluster , copy files from the cluster, install software, submit and monitor jobs etc. We never perform analysis on the login node. If we do the login nodes will choke and other users may not be login or copy files. The analysis are performed on compute nodes and that process we will learn later in the lessons. This is a main difference between executing your program on a remote server and using a cluster to run the program. Lola being a good citizen, acknowledges the importance of using shared resources in a way that will not hinder the other users.

We can view all of the worker nodes with the sinfo command.

[yourUsername@login-2.SAGA ~]$ sinfo
PARTITION AVAIL  TIMELIMIT  NODES  STATE NODELIST
normal*      up 7-00:00:00      1  down* c5-12
normal*      up 7-00:00:00     87    mix c1-[7,13-15,19,22-24,28,31,33,3...........
normal*      up 7-00:00:00    110  alloc c1-[1-6,8-9,11-12,16-18,20-21,............
normal*      up 7-00:00:00      1   idle c1-10
bigmem       up 14-00:00:0      1  drain c6-3
bigmem       up 14-00:00:0     10    mix c3-[29-31,53,56],c6-[1,4-7]
bigmem       up 14-00:00:0      1  alloc c6-2
bigmem       up 14-00:00:0     24   idle c3-[32-52,54-55],c6-8
accel        up 14-00:00:0      3    mix c7-[1-2,8]
accel        up 14-00:00:0      5  alloc c7-[3-7]
optimist     up   infinite      1  down* c5-12
optimist     up   infinite      1  drain c6-3
optimist     up   infinite    100    mix c1-[7,13-15,19,22-24,28,31,33,35,37-38,....
optimist     up   infinite    116  alloc c1-[1-6,8-9,11-12,16-18,20-21,25-27,29-30..
optimist     up   infinite     25   idle c1-10,c3-[32-52,54-55],c6-8

There are also specialized machines used for managing disk storage, user authentication, and other infrastructure-related tasks. Although we do not typically logon to or interact with these machines directly, they enable a number of key features like ensuring our user account and files are available throughout the HPC system.

Shared file systems

This is an important point to remember: files saved on one node (computer) are often available everywhere on the cluster!

What’s in a node?

All of a HPC system’s nodes have the same components as your own laptop or desktop: CPUs (sometimes also called processors or cores), memory (or RAM), and disk space. CPUs are a computer’s tool for actually running programs and calculations. Information about a current task is stored in the computer’s memory. Disk refers to all storage that can be accessed like a file system. This is generally storage that can hold data permanently, i.e. data is still there even if the computer has been restarted.

/hpc-intro/Node%20anatomy

Explore Your Computer

Try to find out the number of CPUs and amount of memory available on your personal computer.

Explore The Head Node

Now we’ll compare the size of your computer with the size of the head node: To see the number of processors, run:

[yourUsername@login-2.SAGA ~]$ nproc --all

How about memory? Try running:

[yourUsername@login-2.SAGA ~]$ free -m

Getting more information

You can get more detailed information on both the processors and memory by using different commands.

For more information on processors use lscpu

[yourUsername@login-2.SAGA ~]$ lscpu

For more information on memory you can look in the /proc/meminfo file:

[yourUsername@login-2.SAGA ~]$ cat /proc/meminfo

Explore a Worker Node

Finally, let’s look at the resources available on the worker nodes where your jobs will actually run. Try running this command to see the name, CPUs and memory available on the worker nodes (the instructors will give you the ID of the compute node to use):

[yourUsername@login-2.SAGA ~]$ sinfo --node c3-12 -o "%n %c %m"

Compare Your Computer, the Head Node and the Worker Node

Compare your laptop’s number of processors and memory with the numbers you see on the cluster head node and worker node. Discuss the differences with your neighbor. What implications do you think the differences might have on running your research work on the different systems and nodes?

Units and Language

A computer’s memory and disk are measured in units called Bytes (one Byte is 8 bits). As today’s files and memory have grown to be large given historic standards, volumes are noted using the SI prefixes. So 1000 Bytes is a Kilobyte (kB), 1000 Kilobytes is a Megabyte, 1000 Megabytes is a Gigabyte etc.

History and common language have however mixed this notation with a different meaning. When people say “Kilobyte”, they mean 1024 Bytes instead. In that spirit, a Megabyte are 1024 Kilobytes. To address this ambiguity, the International System of Quantities standardizes the binary prefixes (with base of 1024) by the prefixes kibi, mibi, gibi, etc. For more details, see here

Differences Between Nodes

Many HPC clusters have a variety of nodes optimized for particular workloads. Some nodes may have larger amount of memory, or specialized resources such as Graphical Processing Units (GPUs).

With all of this in mind, we will now cover how to talk to the cluster’s scheduler, and use it to start running our scripts and programs!

Key Points

  • An HPC system is a set of networked machines.

  • HPC systems typically provides login nodes and a set of worker nodes.

  • The resources found on independent (worker) nodes can vary in volume and type (amount of RAM, processor architecture, availability of network mounted file systems, etc.).

  • Files saved on one node are available on all nodes.