From CompBio
Jump to: navigation, search

This is just a starter for now. Content from the CompBio computing instructions should be imported in here judiciously.


General suggestions

I have no problems with people doing work in any system as they please, but when they come into the compbio cluster, it is these suggestions that are effective the most and following them will make life easy on you. (Rhetorically, why make life difficult for yourself?) These suggestions initially started on June 18, 2011 and replace long standing suggestions going as far back as 1990. Changes have been made to streamline management.

  • One disk/partition per user.
  • Users generally log into the machine physically containing the disk/partition their home directory is on, after logging in from the outside to abyss. (It used to be that each user would get one machine but this was too wasteful since most people work form home and since the farm is used for most of our computing.)
  • Users may work from one of these machines ({up1,up2,...}) or login to one of the other machines below.


  • is the CompBio domain name suffixed to all machine hostnames.
  • {abyss,limbo} represent machines that can be accessed from the outside for logging into compbio.
  • {zen,tao,karma,...} General philosophical themed names reserved for individual users' desktops or laptops.
  • {up1,up2,...} represent the CompBio user workstations/terminals/desktops.
  • {fp1,fp2,...,fp166} represent the CompBio farm proper.
  • {sb1,sb2,...} represent the CompBio sandbox.
  • {sp1,sp2,sp3} represent the CompBio servers.


Most people have a home directory in a central location or a central set of locations. This location is typically a large RAID (RAID-0) volume that is mirrored 100% of the time (so it's automatically backed up), and this is where all the "important" user information is kept. Then the user can have disks spread out across the farm (one disk partition per user) that they can manage for the speed and computation.

Machine set up is so that the system disk contains the system partition(s) and the remaining space is put in a /spare partition. The remaining disks are labelled /maxa, /maxb, ... for each individual disk on a machine. Within a disk, partitions can then be numbered, for example maxa1, maxa2, maxa3, etc. This normally isn't necessary but with individual disk sizes being very big, the old suggestion of person per disk might become obsolete also.



This is of particular use when you set up a new system in our cluster.

  • As you can guess from these naming conventions, usually machines have names ending with numbers >= 1 that correspond to the the IP address of that machine, and disks have names with letters ending with the characters a-z.
  • User chosen names have to signify some philosophy. This is pretty much the only name space left that has not been sterilised.

These names can purport to be a pseudophilosophical reflection of a person/computer.

  • Farm machine names are of the form fpN. fp0 is reserved.
  • Server names are of the form spN. sp0 is reserved.
  • NAS and DAS names are of form: nasN and dasN. nas0 amd das0 is reserved.
  • Individual mount point names are of the form /spare (only for any spare space on the primary hard disk made available as a disk partition), and /maxa, /maxb, /maxc, etc. representing the hard disks available on the machine. Usually most user terminals and farm m

achines will only have the /maxa and /maxb mount points.

  • Other mount point names can take on the form of /dasC for direct attached storage devices where C can be one of a-z.

IP addresses

Should this be public?

  • 10.*.*.* is the CompBio cluster subnet
    • 10.20.*.* is the CompBio local subnet.
      • is the gateway for the local subnet (also firewall and null).
      • is {abyss,limbo}
      •,... (rename amino acid names to {up1,up2,...}
      •,... {printer1,printer2,...}
    • 10.21.*.* is the CompBio DMZ subnet.
    • is the gateway for the DMZ subnet.
Personal tools