Difference between revisions of "Getting Started"

From UFRC
Jump to navigation Jump to search
Line 202: Line 202:
 
UFRC uses the Simple Linux Utility for Resource Management, or '''SLURM''', to allocate resources and schedule jobs. Users can create SLURM job scripts to submit jobs to the system. These scripts can, and should, be modified in order to control several aspects of your job, like resource allocation, email notifications, or an output destination.
 
UFRC uses the Simple Linux Utility for Resource Management, or '''SLURM''', to allocate resources and schedule jobs. Users can create SLURM job scripts to submit jobs to the system. These scripts can, and should, be modified in order to control several aspects of your job, like resource allocation, email notifications, or an output destination.
  
* See the [[Annotated_SLURM_Script|Annotated SLURM Script]] for a walk-through of the basic components of a SLURM job script
+
* See the [[Annotated SLURM Script]] for a walk-through of the basic components of a SLURM job script
* See the [[Sample_SLURM_Scripts|Sample SLURM Scripts]] for several SLURM job script examples
+
* See the [[Sample SLURM Scripts]] for several SLURM job script examples
  
  
Line 215: Line 215:
 
</pre>
 
</pre>
  
View [[SLURM_Commands]] for more useful SLURM commands.
+
View [[SLURM Commands]] for more useful SLURM commands.
  
 
====Managing Cores and Memory====
 
====Managing Cores and Memory====

Revision as of 20:17, 28 July 2022

Welcome to UF Research Computing! This page is intended to help new clients get started on HiPerGator.

From Zero to HiPerGator

Initial Consult

If a face-to-face discussion about the group's needs is needed you can meet one of the UF Research Computing Facilitators face-to-face or virtually or submit a support request to start the conversation.

HiPerGator Accounts

Group's sponsor has to be the first person to request a HiPerGator account indicating that they are a new sponsor. In the process we will create their sponsored group.

Afterwards, group members will be able to submit HiPerGator account requests indicating their PI as the sponsor. Once approved, their linux accounts will be created.

Trial Allocation

We recommend that the group's sponsor request a free trial allocation for storage and computational resources to get the group started on HiPerGator. Group members can then use HiPerGator for the 3 month duration of the trial allocation to figure out what resources and applications they really need.

Purchasing Resources

After or while the group uses a trial allocation to determine the computational and storage resources it needs the group's sponsor can submit a purchase request for hardware (5-years) or services (3-months to longer) to invest into the resources to cover the group's HiPerGator use.

Some groups may have access to shared departmental allocations. In this case, instead of purchasing resources, group members can request to be added to the departmental group to gain access to the shared resources.

Some examples of departments with shared allocations include the Genetics Institute, Emerging Pathogens Institute, Statistics Department, Biostatistics Department, Center for Compressible Multiphase Turbulence (CCMT), Cognitive Aging and Memory Clinical Translational Research Program (CAMCTRP), Center for Molecular Magnetic Quantum Materials, Physics Department, and Plant Pathology Department. In addition, several research groups working on collaborative projects have shared allocations accessible to members of those projects.

At this point a group is established on HiPerGator and can continue their computational work. See below for more details on the basic use.

Introduction to Using HiPerGator

To use HiPerGator or HiPerGator-AI you need three basic parts

  • Interfaces

You use Interfaces to interact with the system, manage data, initialize computation, and view the results. The main categories of interfaces 'Command-Line' also known as Terminal, Graphical User Interfaces, and Web Interfaces or applications for more specialized use. Some distinctions here are blurred because, for example, you can open a Terminal while using a Web Interface like JupyterHub or Open OnDemand, but mostly you use a command-line Terminal interface through SSH connections (see below).

  • Data Management

To perform research analyses you need to upload and manage data. Note that misuse of the storage systems is the second main reason for account suspension after running analyses on login nodes.

  • Computation

Warning: do not run full-scale (normal) analyses on login nodes. Development and Testing is required reading. The main approach to run computational analyses is through writing job scripts and sending them to the scheduler to run. Some interfaces like Open OnDemand, JupyterHub, and Galaxy can manage job scheduling behind the scenes and may be more convenient than job submission from the command-line when appropriate.

Interfaces

Connecting to a HiPerGator Terminal via SSH

To work on HiPerGator you will have to connect to hpg.rc.ufl.edu from your local computer either via SSH (terminal session) or via one of the web/application interfaces we provide such as Galaxy, Open OnDemand, or JupyterHub. The default port 22 will require password authentication whereas port 2222 can be used for ssh key authentication.

Note that the UF security requirements include the use of Multi-Factor Authentication when connecting to HiPerGator. See the SSH Multiplexing help if you are connecting from Linux or MacOS to reduce the frequency of MFA use.

For any given command below, <username> should be replaced with the UFRC username (same as your GatorLink username).

Connecting from Windows

Expand this section to view instructions for logging in with Windows.

Since Microsoft Windows does not come with a built-in SSH client, you must download a client from the web.

For University-managed computers PuTTY, Tabby, and Git Bash are approved for 'fast track' installations.

PuTTY

  • Download PuTTY to your local machine and start the program.
  • Connect to hpg.rc.ufl.edu.
  • At the login prompt, enter your username (this should be the same as your GatorLink username)
  • Enter your password when prompted. You are now connected and ready to work!

Tabby

  • Download Tabby to your local machine: tabby-version#-setup.exe or tabby-version#-portable.zip for a portable version.
  • Star the program and click Settings > Profiles > +New profile > SSH connection
 Name: HiPerGator
 Host: hpg.rc.ufl.edu
 Username: <username>
 Password: "Set password" or "Add a private key"
  • Click "Save"
  • Click on the window icon "New tab with profile" and select "HiPerGator hpg.rc.ufl.edu"
  • You are now connected and ready to work!

Connecting from Linux and MacOS

Expand to view instructions for connecting from Linux or MacOS.

You may omit GatorLinkUsername if the username on your local computer matches your GatorLink username e.g.

Open a terminal and run

$ ssh <GatorLink Username>@hpg.rc.ufl.edu
or
$ ssh hpg.rc.ufl.edu

Enter your password when the prompt appears. You are now connected to a login node and your working directory will be /home/<GatorLink>, ready to work!

Logging in via web browser

Expand to view instructions for Logging in via web browser.

  • Services like OpenOnDemand (OOD) and JupyterHub require the users to be on the UF network. If the user is off the campus network they need to be connected to the UF VPN.
  • To start a GUI application go to https://ood.rc.ufl.edu and choose the application you would like to use, for example HiPerGator Desktop is a linux GUI.
    • To start a Jupyter Notebook with OOD, select the “Jupyter Notebook” application server under the “Interactive Applications” tab on the main dashboard. Specify the resources you need in the resource request and click “Launch” at the bottom of the form. Once your job starts, click “Connect to Jupyter” to start your Notebook.
  • To start a Jupyter Notebook go to http://jhub.rc.ufl.edu and select the resources you need. This JupyterHub server has preset resource options.
    • Jupyterhub and OutOfMemory (OOM) – if your Notebook stops responding it might be an indication of the job running out of memory. Currently this does not produce an error message. You can check your /home/<user>/jupyter slurm spawner logs to be sure. You will need to restart the job with additional RAM.
  • To start a Galaxy instance go to https://galaxy.rc.ufl.edu/. Galaxy is a web-based framework for accessible, reproducible, and transparent biological computing.

Invalid Password

If you have forgotten your password or need to change it, use the links below to set a new one:

Change or Reset Password

NOTE: Research Computing now uses Gatorlink Authentication for all passwords. Please use the UF Gatorlink password system for all password issues.

Data Management

Transferring Data

If you need to transfer datasets to or from HiPerGator and your local computer or another external location you have to pick the appropriate transfer mechanism.

  • For small or medium file transfers use sftp, scp, or rsync to login to sftp.rc.ufl.edu, hpg.rc.ufl.edu, or rsync.rc.ufl.edu.
  • For large file transfers or transfers with many small files, use our Globus service.

For more in-depth information see Transfer Data.

SFTP

SFTP, or secure file transfer, works well for small to medium data transfers and is appropriate for both small and large data files.

If you would like to use a Graphical User Interface secure file transfer client we recommend:

After you have chosen and downloaded a client, configure the client to connect to hpg.rc.ufl.edu, specifying port number 22. Use your username and password to log in.

Rsync

If you prefer to use the command-line or to get maximum efficiency from your data transfers Rsync, which is an incremental file transfer utility that minimizes network usage, is a good choice. It does so by transmitting only the differences between local and remote files rather than transmitting complete files every time a sync is run as SFTP does. Rsync is best used for tasks like synchronizing files stored across multiple subdirectories, or updating large data sets. It works well both for small and large files. See the Rsync page for instructions on using rsync.

Globus

Globus is a high-performance mechanism for file transfer. Globus works especially well for transferring large files or data sets

Samba

Samba service, also known as a 'network share' or 'mapped drive' provides you with an ability to connect to some HiPerGator filesystems as locally mapped drives (or mount points on Linux or MacOS X).Once you connected to a share this mechanism provides you with a file transfer option that allows you to use your client computer's native file manager to access and manage your files. UFRC Samba setup does not provide high performance, so try to use it sparingly and for smaller files, like job scripts or analysis reports to be copied to your local system. You must be connected to the UF network (either on-campus or through the VPN) to connect to Samba shares.

Automounted Paths

Note: NFS-based storage on our systems are typically automounted, which means they are dynamically mounted only when users are actually accessing them. For example if you have an invested folder as /orange/smith, to access it you will have to specifically type in the full path of "/orange/smith" to be able to see the contents and access them. Directly browsing /orange will not show the smith sub-folder unless someone else is using it coincidentally. Automounted folders are pretty common on the systems, they include /orange, /bio, /rlts and even /home etc.

Editing your files

Several methods exist for editing your files on the cluster.

Native Editors

  • vi - The visual editor (vi) is the traditional Unix editor; however, it is not necessarily the most intuitive editor. View a tutorial for using vi
  • emacs - Emacs is a much heavier duty editor, but again has the problem of having commands that are non-intuitive. View a tutorial for using emacs
  • nano - Nano has a good bit of on-screen help to make it easier to use.

External Editors

You can also use your favorite file editor on your local machine, and then transfer the files to the cluster afterward. A caveat to this is that files created on Windows machines usually contain unprintable characters, which may be misinterpreted by Linux command interpreters (shells). If this happens, there is a utility called dos2unix that you can use to convert the text file from DOS/Windows formatting to Linux formatting.

Computation

HiPerGator Etiquette

  • Only run workloads on compute nodes. Do not run scripts or applications on the login nodes beyond a small quick test. Use sbatch, srundev, salloc, or srun to start a session on a compute node instead.
  • Only run workloads from blue storage. This is a fast storage systems that can handle the I/O involved in research workloads. Before using sbatch or launching a workload interactively, make sure your working directory is a blue file path, e.g. /blue/<group>/<user>, and not your home directory (~ or /home/<user>). Use pwd to print working directory.
  • Home directory is only for storing user readable files. Your 40GB home directory is the only storage on HiPerGator for which file recovery may be possible, so keep copies of scripts, configurations, or other important files here. Versions of files may be available for the previous 7 days (see information on home directory snapshots and recovering files), if you need to recover a file from your home directory. If you need back ups of important data, backup services will need to be purchased.
  • Do not install new software when using existing modules. This will cause errors when using our software environments, because the new installation is in your local folder. If you need to install software, create a Conda virtual environment, or open a support ticket to request RC to install the software in one of our software environments if it will be widely used.
  • Do not run workloads from orange storage. Orange is intended as long-term, archival storage of data you currently do not use. It cannot handle the high-throughput requirements of high-performance computing workloads.
  • Do not request excessive resources. This includes CPU, GPU and memory. Job emails include summary estimations of memory use, however, active monitoring will help you understand resource requirements. Applications often require special commands, arguments, or configurations to run in parallel. Therefore, you will likely need to do more than request multiple CPUs or GPUs for a workload to put those resources to use.

Using installed software

The full list of software available for use can be viewed on the Installed Software page. Access to installed software is provided through Environment Modules.

The following command can be used to browse the full list of available modules, along with short descriptions of the applications they make available:

module spider

To load a module, use the following command:

module load <module_name>

In Jupyter Notebooks, kernels are available with our most popular software stacks. If you are unable to find what you need or would like software installed, please fill out a help request.


For more information on loading modules to access software, view the page on the basic usage of environment modules.

There are some useful commands and utilities in a 'ufrc' environment module in addition to installed applications.

Interactive Testing or Development

You don't always have to use the SLURM scheduler. When all you need is a quick shell session to run a command or two, write and/or test a job script, or compile some code use SLURM Dev Sessions.

Running Graphical Programs

It is possible to run programs that use a graphical user interface (GUI) on the system. However, doing so requires an installation of and configuration of additional software on the client computer.

Please see the GUI Programs page for information on running graphical user interface applications at UFRC.

Scheduling computational jobs

UFRC uses the Simple Linux Utility for Resource Management, or SLURM, to allocate resources and schedule jobs. Users can create SLURM job scripts to submit jobs to the system. These scripts can, and should, be modified in order to control several aspects of your job, like resource allocation, email notifications, or an output destination.


To submit a job script from one of the login nodes accessed via hpg.rc.ufl.edu, use the following command:

$ sbatch <your_job_script>

To check the status of submitted jobs, use the following command:

$ squeue -u <username>

View SLURM Commands for more useful SLURM commands.

Managing Cores and Memory

See Account and QOS limits under SLURM for the main documentation on efficient management of computational resources.

The amount of resources within an investment is calculated in NCU (Normalized Computing Units), which correspond to 1 CPU core and about 3.5GB of memory for each NCU purchased. CPUs (cores) and RAM are allocated to jobs independently as requested by your job script.

Your group's investment can run out of **cores** (SLURM may show QOSGrpCpuLimit in the reason a job is pending) OR **memory** (SLURM may show QOSGrpMemLimit in the reason a job is pending) depending on current use by running jobs. The majority of HiPerGator nodes have the same ratio of about 4 GB of RAM per core, which, after accounting for the operating system and system services, leaves about 3.5 GB usable for jobs; hence the ratio of 1 core and 3.5GB RAM per NCU.

Most HiPerGator nodes have 32 cores and 128 GB RAM (~30,000 cores in the newer part of the cluster) or 64 cores and 256 GB RAM (~16,000 cores in the older part of the cluster). The bigmem nodes and the newer Skylake nodes have a higher ratio of 16 GB/core and 6 GB/core, respectively. See Available_Node_Features for the exact data on resources available on all types of nodes on HiPerGator.

You must specify both the number of cores and the amount of RAM needed in the job script for SLURM with the --mem (total job memory) or --mem-per-cpu (per-core memory) options. Otherwise, the job will be assigned the default 600mb of memory.

If you need more than 128 GB of RAM, you can only run on the older nodes, which have 256 GB of RAM, or on the bigmem nodes, which have up to 1.5 TB of RAM.

See Account and QOS limits under SLURM for an extensive explanation of QOS and SLURM account use.

Monitoring Your Workloads

You can see presently running workloads with the squeue command e.g.

$ squeuemine

OpenOnDemand offers a method to monitor jobs using the Jobs menu in the upper toolbar on your dashboard. This will show your current running, pending, and recently completed jobs. Select: Jobs -> Active Jobs from the upper dashboard menu.


We provide a number of helpful commands in the UFRC module. The ufrc module is loaded by default at login, but you can also load the ufrc module with the following command:

$ module load ufrc


Examples of commands for SLURM or HiPerGator specific UFRC environment module

Expand this section to view Examples of commands

$ slurmInfo           - displays resource usage for your group
$ slurmInfo -p        - displays resource usage per partition
$ showQos             - displays your available QoS
$ home_quota          - displays your /home quota
$ blue_quota          - displays your /blue quota
$ orange_quota        - displays your /orange quota
$ sacct               - displays job id and state of your recent workloads
$ nodeInfo            - displays partitions by node types, showing total RAM and other features
$ sinfo -p partition  - displays the status of nodes in a partition
$ jobhtop             - displays resource usage for jobs
$ jobnvtop            - displays resource usage for GPU jobs
$ which python        - displays path to the Python install of the environment modules you have loaded

Getting help

If you are having problems using the UFRC system, try our resources at Help or let our staff know by submitting a support request.

Visual Overview

The diagram below shown a high-level overview of HiPerGator use. We will go over each part in sections below

HiPerGator.png