---
tags: User
---

<a href="https://hackmd.io/@teoroo-cluster" alt="Teoroo Cluster">
<img src="https://img.shields.io/badge/Teoroo--CMC-All%20Public%20Notes-black?style=flat-square" /></a>

# Teoroo cluster: hardware

## Overview

The TEOROO cluster consists of two parts. The old cluster: `teoroo`
is planned to be decommissioned from 2019. The new cluster `teoroo2`
is online in 2018. Currently the compute nodes
in teoroo is moved to teoroo2. The domain teoroo.kemi.uu.se
will only be used to serve the group web page in the future.

## teoroo2

List of machines in the cluster:
- **router**: NIS, DHCP and DNS server[^router]
- **clooney**: NFS server
- **brosnan**: login node and GPU compute node
- **jackie**: GPU compute node
- **aberlour**: old 16 TB RAID6 NFS server
- **w1-w8**: old compute nodes[^w6down]

**Note: on the GPU nodes brosnan and jackie there are SSD disks `/scratch2/` that
you can use for io-intensive tasks, such as fast access to some frequently used
datasets.**

### Specifications

| Machine | Spec.[^cpucores] | System |
| ------------ | --------------------------------------- | --------------------- |
| **router** | Opteron 2356 (4x2 cores); 8GB Mem. | Ubuntu 16.04 |
| **clooney** | Xeon E5-2630 v4 (10 cores); 128 GB Mem. | Ubuntu 16.04 |
| **brosnan** | Xeon E5-2695 (18x2 cores); 256GB Mem. | Ubuntu 16.04 |
| **jackie** | Xeon E5-2695 (18x2 cores); 256GB Mem. | Ubuntu 16.04 |
| **aberlour** | Xeon E5620 (4 cores); 8GB Mem. | Scientific Linux 6.10 |
| **w1-w3** | Xeon X5675 (6x2 cores); 24GB Mem. | Ubuntu 16.04 |
| **w4-w8** | Xeon X5675 (6x2 cores); 24GB Mem. | Scientific Linux 6.10 |

## teoroo

Below is some information about `teoroo` the old cluster.


- **Lochlomond**: very old comp with PATA HDDs - powered off
- **ppc**: re-purposed as router for the new cluster
- **Allium**: broken - remains located in the 4^th^ floor
- **studentlogin**: login node for the old student cluster
- **Mackmyra**: 4TB RAID6, NFS, NIS, DHCP and DNS server
- **Aberlour**: 16 TB RAID6, NFS server (moved to the new cluster)
- **w1-w8**: CPU compute nodes (moved to the new cluster)

[^router]: not accessible for the users
[^w6down]: w6 is temporarily down
[^cpucores]: Here listed are the (no. of cores per socket) x (no. of sockets); most of the machines are hyperthreaded so the actual thread count might be doubled, though computational intensive tasks typically do not benefit from no. of threads beyond the number of physical cores.