How to set up a virtual machine on the Nectar cloud
What is it
Set up some stuff in: Dashboard -> Project -> Compute
- create key pair - called it nectarkeys
- click on it. it should download the private key with a .pem file type
- put it in your .ssh folder
- change permissions for this file: chmod u+rw,og-rwx nectarkeys.pem
- launch instance
- name it
- then click on the source tab in the left hand panel
- find e.g. NeCTAR Ubuntu 16.04 LTS (Xenial) amd64, and tick
- flavour tab: eg m3.xsmall
- note: Small (1 core), Medium (2 cores), Large (4 cores), Extra-large or XL (8 cores), XXL (16 cores)
- security groups tab: click on both default and ssh
- key pair tab: make sure my “nectarkeys” is showing
- bottom right: launch instance
Your instance (= VM)
- should now be running; might take a few minutes to show up.
- find IP address e.g. 115.333.444.255.
- This identifies this particular instance.
- Look at the Flavour column and click on the flavour (e.g. m3.xsmall).
- This shows how many cpus (e.g. 1) and how much RAM (e.g. 2GB)
We will login (= connect) to this instance (= VM).
Work out your username
- open terminal (or equivalent)
- type in:
ssh -i .ssh/nectarkeys.pem firstname.lastname@example.org // replace with your info
-i // flag for key file
.ssh/nectarkeys.pem // key file location (.ssh/) and name (nectarkeys.pem)
ubuntu // username for this VM
15.333.444.255 //IP address for this VM
Note: This is about getting tools (software) onto your VM. There are other ways to do this!
- While still logged in to your VM:
- Need to disconnect (type in exit) then re-connect (with
ssh and your info)
- Type in these lines one at a time:
conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge
e.g. the spades assembler
Get some sample data
e.g. some Illumina fastQ sequencing reads
Run an analysis
e.g. spades assembly. Note that we will set number of threads (1) and max amount of RAM (2 GB), as per the VM’s limits.
How to close and keep it running
- If you shut your terminal or lose your internet connection, the run will stop.
- To be able to shut and have it keep running, you can add some extra commands before and after the usual commands:
nohup [usual commands here] 2>&1 &
nohup at the start (means no hangup)
2>&1 after the commands (means error messages will go to the same place as output info)
& at the end of the line (puts the job to run in the background)
- You can see the job running if you type in
top (q to exit)
- While this runs in the background, you can still work in the terminal window, or exit.
- When you re-login, you’ll see any output files (if it’s finished), plus a file called nohup.out which has the info that would normally be printed to the screen during the run.