#####Add a Group Resource Profile for Comet (GRP)

1. What is a group resource profile?
   • Group resource profile is a configuration which states how a particular HPC resource (campus, national, cloud) is available for users to use.
   • One or many HPCs can be added to a GRP and then it can be shared with the users and user groups to enable use.
   • A gateway can have multiple GRPs and a gateway user can be in one or many GRPs
   • At the time of experiment creation, user need to select which GRP to move forward to launch experiment jobs.
2. To Create a GRP
   • Navigate to Settings → Group Resource Profile.
   • Click New Group Resource Profile +.
   • Add Name: Default Gateway Profile
   • Default SSH Credential: Select a key from the list or generate a new one.
   • Click New Compute Resource +.
   • Select the resource from the list provided.
     • In next page provide the details required for SSH communication
     • Login Username: This is the login name to use in SSH login
     • SSH Credential: Select from the list, or generate a new one
     • Allocation Project Number: this is applicable for resources which maintains an account for user allocation and usage.
     • Scratch Location: The location for users' computational working directories
   • Provide computational queue policies
     • This section will provide details of maximum node, walltime and CPUs per HPC resource that a gateway user could use in experiment creation.
     • If you don't specify these limits, then the maximum allowed from the HPC resource will be available for users.
     • NOTE: This feature is more popular with gateways which are used in educational purposes and also for gateways where the gateway admin provides XSEDE resoruces for free via community allocations.
     • Click Save
     • Next page, click Save

Register your Compute & Storage Resources

NOTE: Only SciGaP admins or gateway service provider can add compute resources and storage resources.
If the gateway and middleware is hsoted by you, You can add the Compute Resources and Storage Resource.
What is a compute resource?
A compute resource is an object that represents a host, host cluster, or pool in a virtualization platform, a virtual data center, or an Amazon region on which machines can be provisioned.
Compute resource could come as national resources, campus resources or even private cluster.
NOTE:For testing purposes users (mainly gateway developers) this could even be the local machine.
What is a storage resource?
A storage resource is the server that stores all user data files: application input files and output files.
Most of the time this is the same server that the gateway is deployed in. The storage resource can be either provided by the gateway provider or by you.
If the gateway service provider is providing the resource, then there will be file retention policies in place.

Compute Resource Information

Provide the information below to get your compute resource/HPC registered with Super admin portal, SciGaP

1. HPC Name (This is the name used to SSH to the resource)
   
2. Queue details (Queue name, maximum limits of nodes, CPUs and Walltime.)
   
3. Resource manager type (SLURM, PBS)
   
4. Job command binary path
   

Register a XSEDE Resource

NOTE: Please note this set of instructions are ONLY for gateways hosted by the user. If not SciGaP admins will do this configuration for you in SciGaP Portal.

1. Navigate to Admin Dashboard → Compute Resource → Register
   
2. Provide
   • Host Name: comet.sdsc.edu
     
   • Host Aliases:
     
   • IP Addresses:
     
   • Resource Description: SDSC Comet Cluster ( Just a description)
     
   • Create
     
3. In Queues tab
   
   • Click ‘Add a Queue’
     
   • Queue 1
     • Queue Name: compute ( Per resource this is a unique value. Once created cannot edit the name. If need to change, delete and create again)
       
     • Queue Description: Used for access to regular compute nodes
       
     • Queue Max Run Time (In Minutes): 280 (Note required for the local resource)
       
     • Queue Max Nodes: 72 (Note required for the local resource)
       
     • Queue Max Processors: 1728 (Note required for the local resource)
       
     • Max Jobs in Queue: 0 (Note required for the local resource)
       
     • Max Memory For Queue( In MB ): (Note required for the local resource)
       
   • Queue 2
     • Queue Name: shared ( Per resource this is a unique value. Once created cannot edit the name. If need to change, delete and create again)
       
     • Queue Description: Single-node jobs using fewer than 24 cores
       
     • Queue Max Run Time (In Minutes): 2880 (Note required for the local resource)
       
     • Queue Max Nodes: 1 (Note required for the local resource)
       
     • Queue Max Processors: 24 (Note required for the local resource)
       
     • Max Jobs in Queue: 0 (Note required for the local resource)
       
     • Max Memory For Queue( In MB ): (Note required for the local resource)
       
4. File System
   
   • Nothing to add here, this is futuristic development.
     
5. Job Submission Interface
   
   • Click on the tab
     
   • Add a new Job Submission Interface
     
   • Job Submission Protocol: SSH
     
   • Select Security Protocol: SSH_KEYS
   • SSH Port: 22
   • Select resource manager type: SLURM
     
   • Job Manager Bin Path: /usr/bin/
   • Job Commands (SUBMISSION): sbatch
     
   • Job Commands (JOB_MONITORING): squeue
     
   • Job Commands (DELETION): scancel
     
   • Parallelism Prefixes (MPI): mpiexec
     
   • Add Job Submission Protocol
     
6. Data Movement Interface
   
   • Add a new Data Submission Interface
     
   • Data Movement Protocol: SCP
     
   • Select Security Protocol: SSH_KEYS
     
   • SSH Port: 22
     
   • Add Data Movement Protocol
     Now the Local resource is ready for job submissions.
     Comet is ready for job submissions.

Register a Campus Resource

NOTE: Adding a campus resource is similar to adding a XSEDE resource. Same steps to follow with similar information.