|
|
# Quick start guide
|
|
|
|
|
|
## How to connect
|
|
|
|
|
|
Use the ssh protocol and connect through shark or the ssh jump server with your research username and password.
|
|
|
|
|
|
Slurm shark ip addres login node: 145.88.65.251
|
|
|
|
|
|
from here you have 2 options:
|
|
|
|
|
|
interactive login with : srun
|
|
|
batch submission with : sbatch
|
|
|
|
|
|
### Directories
|
|
|
The directories /home, /exports, /bam-export are all the same as Shark.
|
|
|
Please keep in mind that some programs create .config files inside your own home directory.
|
|
|
This can cause binaries, pipelines etc. to use a wrong config file.
|
|
|
|
|
|
The scrips directive for slurm is #SBATCH, you will need to rewrite your submission scripts to work with slurm.
|
|
|
|
|
|
### cgroups
|
|
|
cgroups (control groups) limits, isolates and measures resource usage of a group of processes.
|
|
|
SLURM works together with cgroups to limit your resources, what you get by default or have asked for iswhat you can only use.
|
|
|
Lets say you ask for 8G mem with the option --mem=8G and with 2 cores with --ntasks-per-node=2,
|
|
|
when this jobs starts running cgroups will ensure that the only resources you will have available is 2 cores and 8G mem.
|
|
|
|
|
|
### Default limits
|
|
|
* cores 1
|
|
|
* memory 2G
|
|
|
* run time 1 hour
|
|
|
|
|
|
|