Mek-quake is a cluster of dual-core dual-processor Opteron servers. Originally there were 65 nodes, but only 38 remain. The CPUs are Opteron 280s (2.4GHz clock speed) and each machine has 4Gb of RAM. They all run Linux.
Mek-quake can only be used by sshing into the head node, whose external name is mek-quake.ch.cam.ac.uk. Almost all work is done from there. In particular, passwords and shells should only be changed on the head node. Every node has a name on the cluster's internal network of the form compute-X-Y . Unlike some other local clusters, you cannot log into compute nodes unless you have been assigned that node by the queueing system.
Homespace is on a disk array attached to the head node. The /home filesystem is about 200Gb in size. This is shared between all nodes so you see the same home directory wherever you (or your job) are on the machine. /home is backed up nightly and two weeks of incrementals are kept. There are quotas on /home. Currently the soft limit is 6Gb and the hard limit is 8Gb.
You will also have a directory under /sharedscratch. Data stored here is not backed up. Quotas on that data are set at 70Gb soft and 90Gb hard. At the moment I have no plans to purge it regularly, so please clean up old files when you're done with them. Please try to use /sharedscratch appropriately: it should be used for data that you could recreate in a reasonable time. Each node also has a local /scratch filesystem on which you will have your own directory once you've run a job on the node. These filesystems are about 120Gb in size with no quota restriction. They are local to each node and so considerably faster than the NFS-mounted /home and /sharedscratch. Please clean up files on /scratch when you are done with them. As you cannot log into the nodes directly, the scratch directories can be accessed under /nodescratch/compute-X-Y on the head node. This is controlled by an automounter so the directories are only mounted when you reference them; ie you must explicitly type something like cd /nodescratch/compute-0-0 to make the filesystem appear.
The system has third party software packages installed in /usr/local. In order to manage all the combinations of compilers and libraries, the modules environment is installed. By default the latest 64-bit compilers and OpenMPI are loaded.
The head node also has some extra software packages, including popular editors. If there is a package missing from the head node that you would like to use then please ask; it will probably be possible to install it provided it is a sensible size. However compute work should not be run on the head node.
The parallel environments on the system are dealt with in a separate document .
All compute jobs must be run through the queueing system. The queueing system will assign a number of nodes to you and run your script on the first node, copying the output back to a user-specified file at the end of the job. The queueing system is SLURM. Read the SLURM documentation for how to use SLURM.
There is one node (compute-1-3) which has 8GB RAM for running larger memory jobs. Use the usual SLURM commands to request it.
Problems with Mek-quake should be reported to <firstname.lastname@example.org> in the first instance.