Connecting to CGRL Vector cluster - melisakman/Helianthus GitHub Wiki

The first thing you would like to do is to request access to CGRL. Here is the form to do that.

But first simply send an e-mail to Jason Huff (CGRL coordinator) at [email protected].

After you get your log-in details, follow the steps here. You need to install the google authenticator, a password generator that will allow you to connect to the cluster through ssh. Once you have that going open a terminal window and ssh by:

$ ssh [email protected]

You will be prompted for a password and now you enter your password followed by the 6-digit number created by google authenticator. I am afraid you need to do this every time you ssh to the cluster as well as when you are transferring files through scp (and we will talk about that too). You should be hooked to CGRL after you enter your password.

You now should be in your home folder: /global/home/users/UserName. This folder has a limit of 10 GB an is not for data storage. I keep my scripts and some results that are small in size. This folder is backed up. One good practice is to have a git folder here. I usually create my slurm scripts on my local computer and then synch my git folder on my local computer with a git folder on my home folder. If you wonder how to use git, look at the git section on the wiki.

You can check your folder by Print Working Directory:

$ pwd

Now let's have a look at some basics on directories and common Blackman folders.

1. vector scratch is a place to store data that will be analyzed on vector cluster. And you have your own folder on this scratch here: /clusterfs/vector/scratch/UserName You need to be careful about this scratch space since it fills quite easily (although I think now we have more space available) and lead to crushing jobs as there is no space left to write your results. So check how much space is available before you start a job that needs a lot of space. For this simply do:

$ cd /clusterfs/vector $ df -h If you think there is enough space for your analyses, go for it. 2. savio scratch space is a place to store data that will be analyzed on savio. Here is that folder of yours: /global/scratch/UserName

3. There is a shared folder that all of the Blackman Lab can use. This space is 300 GB and is a good place to store things that can be useful to all of us such as blast databases, genome sequences etc. Here we have 2 main folders Mimulus and Helianthus. Please use these folders wisely and do not fill up the shared space. Here is that folder:

$ cd /clusterfs/vector/instrumentData/blackmanlab/ and check out the folders: $ ls