UCR RC Nautilus Cluster Tools

Scripts and configuration files developed and maintained UCR Research Computing that enables easy interaction with the Pacific Research Platform's Nautilus Cluster for UCR Researchers. Helping to promote PRP efforts.# UCR RC Nautilus Research Nodes

For UCR Researchers to provision and access Nautilus Cluster Research Nodes. These are very powerful GPU-enabled virtual servers made available by the Pacific Research Platform with access made possible and facilitated by UCR Research Computing.

Kubernetes

The PRP is a Kubernetes cluster and therefore requires the use of a Kubernetes interface. The command line Kubernetes interface kubectl is what is used in this workflow.

Install

To run these actions locally on your local laptop/workstation you will need to install kubectl with the following commands:.

You can install kubectl via conda:

conda create -n kube -c anaconda-platform kubectl

If you do not yet have conda installed, follow these instructions.

You can install kubectl manually:

curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
mv kubectl ~/bin/.

Config

For kubectl to function, it requires your config file provdied by PRP. In order to get the PRP Kubernetes config file, do the following:

1. Visit Nautilus Portal
2. Click on Login in upper right coner.
3. Login using CILogon credentials (UCR netID).
4. Once authenticated, click on the Get config in the upper right conner.
5. This takes a while to dynamically generate, just wait and eventually your browser will present you a download prompt.
6. Place this file in your ~/.kube directory.

Next set the namespace, or else you will have to append the -n ucr-rc flag to every Kubernetes command:

kubectl config set-context nautilus --namespace=ucr-rc

Nautilus script

Run with out any arguments and this script will spawn a RC Nautilus Research Node and drop you into the shell. You can also run abritraty commands on pop up workstations and list any active workstations you might have running. Please see the help for the command below for details on what options you can and funtions you can do. You can spawn any container you find on docker hub or other repos sas well as the default UCR Research Computing Nautilus Research Node.

./nautilus -h
usage: nautilus [-h] [-p CPUS] [-m MEMORY] [-g GPUS] [-n NAME] [-a AVAIL]
                [-r RUNNING] [-i INTERACTIVE] [-c CONTAINER] [-d COMMAND]

optional arguments:
  -h, --help            show this help message and exit

required named arguments:
  -p CPUS, --cpus CPUS  How many CPUs requested.
  -m MEMORY, --memory MEMORY
                        Number of GB of memory requested.
  -g GPUS, --gpus GPUS  Number of GPUs requested.
  -n NAME, --name NAME  Name for your workstation.
  -a AVAIL, --avail AVAIL
                        List common avaiable containers.
  -r RUNNING, --running RUNNING
                        List running workstations.
  -i INTERACTIVE, --interactive INTERACTIVE
                        To enter interactive mode.
  -c CONTAINER, --container CONTAINER
                        Name of container from DockerHub or PRP GitLab.
  -d COMMAND, --command COMMAND
                        Command to run on the workstation if not a shell.

Launch node using the nautilus script.

./nautilus
Launching forsythc-research-lab-8159 Workstation Node ...
(base) jovyan@forsythc-research-lab-8159--1-jtvlx:~$

This drops you into the shell and you can install and use the node how ever you like.

You can spawn any container you find on docker hub and other repos

./nautilus -c ubuntu:latest
./nautilus -c centos:latest
./nautilus -c rockylinux/rockylinux:latest
./nautilus -c gitlab-registry.nrp-nautilus.io/ucr-research-computing/ucr-rc-nautilus-genomics-conda-node:latest

You can run arbitrary command on any container

./nautilus -c ubuntu:latest -d "uptime"
./nautilus -c centos:latest -d "cd /sharedvol ; wget https://ftp.ncbi.nlm.nih.gov/10GB ; ls"
./nautilus -c rockylinux/rockylinux:latest -d "hostname"
./nautilus -c gitlab-registry.nrp-nautilus.io/ucr-research-computing/ucr-rc-nautilus-genomics-conda-node:latest -d "conda install pandas"

Reconnecting to a node or deleteing a node

When you exit the script it will ask if you want to save it for later use.

./nautilus
Launching forsythc-research-lab-8159 Workstation Node ...

(base) jovyan@forsythc-research-lab-8159--1-jtvlx:~$ exit
Enter 'y' to keep the forsythc-research-lab-8159 workstation node running for later?y
ok

Reconnect to the node
-- 'kubectl exec -it ucr-rc-nautilus-research-node-forsythc--1-njq66 -- bash'

Copy smaller files to the node
-- 'kubectl cp LOCALFILES ucr-rc-nautilus-research-node-forsythc--1-njq66:.'

Copy smaller files from the node
-- 'kubectl cp ucr-rc-nautilus-research-node-forsythc--1-njq66:REMOTEFILES LOCALFILES'

Delete the node
-- 'kubectl delete job ucr-rc-nautilus-research-node-forsythc'

Research Nodename:
-- ucr-rc-nautilus-research-node-forsythc--1-njq66

-------------------------------------------------------

This will also give you the workstation's name (nodename)

kubectl get jobs
kubectl get pods

NAME                         COMPLETIONS   DURATION   AGE
forsythc-research-lab-1008   0/1           64s        64s

In the exmaple above the Research Workstation Name (nodename/jobname) = forsythc-research-lab-1008

Connect to Research Node

kubectl exec -it nodename -- bash

Upload data

kubectl cp localfile nodename:.

Download data

kubectl cp nodename:./remotefile .

Delete Research Node

kubectl delete jobs forsythc-research-lab-1008

Trouble

From pod list, check log:

kubectl logs nodename 

Jobs and pods will expire after 1 week, however you can alter this with the following:

ttlSecondsAfterFinished=604800

Check on the Research Node details:

kubectl describe pod nodename 

Links

• Help - https://element.nrp-nautilus.io
• Resources - https://nautilus.optiputer.net/resources
• Monitoring - https://grafana.nautilus.optiputer.net