White dot for spacing only
The Dice Project

Machine Deliveries

All bulk orders will be delivered to JCMB. This will allow the details of the equipment, both base units and monitors, to be entered into the inventory database by support staff at JCMB. Equipment will not be allocated until this process is complete.

Smaller orders should be delivered to the requesting site for the attention of the Site Liaison Officer or support staff. All invoices and supporting documentation (e.g. delivery notes) should be sent to Sheila for payment and entry into the inventory database. Site Liaison officers may also wish to keep a photocopy of any documentation.

Machine Allocations

A policy document on machine allocation can be found here.

It should be the responsibility of support staff to record the whereabouts of unallocated machines. All machines, before being allocated, should be given a profile with the unallocated.h header file to make tracking easier.

On allocation, the profile should contain the correct inv.location and inv.allocated field with the full name of the person. Any other useful information e.g. on loan should appear in the inv.comment field. This is how we should keep track of machines that are on loan until grant funding is sorted out. A date should be added to the comment field so that we can track machines that should be recovered.

Machine Transfers between Sites

When machines need to be moved between sites, a ticket should be created in the techs queue. We should give at least 3 days notice for a small delivery but for bulk deliveries (more than 1 van load), we should give at least 1 week's notice.

Hot Spares

All sites should have a number of hot spares. We need to be able to minimise disruption to users when a machine is faulty but also balance this against having a store of unused machines. We also frequently have to react quickly to requests for machines for new staff.

AT and JCMB should start with a minimum of 5 spare. AT, in particular, has become very busy so pulling machines out of labs to use as spares should be avoided if possible. With there being no lab at BP from which to pull machines, there should also be an initial pool of 5 machines. FH possibly doesn't require as many spares. The lab in A20 is under-used so spares could readily be pulled from there. However, the machines in A20 are not of a particularly high spec. and therefore could only be used in the very short term for replacement staff machines. It would therefore make sense to also start with 5 spare machines at FH.

We should then monitor allocation of these machines and ensure that no site falls below 2. We could ship machines around the sites as required and only consider buying more when we fall below 10 in total across sites. This would help to prevent a stock of unused machines building up.

To make best use of these hot spares, we should have them installed and added to the condor pool. Apart from providing more compute power, it also has the advantage that when we come to pull one out for 'mainstream' use, it will have all the recent updates.

Replacement Machines.

CEG will decide on the replacement cycle but it will be the responsibility of the frontline support manager to identify machines that are due to be replaced and report this information back to CEG.

 : Units : User_support 

Mini Informatics Logo - Link to Main Informatics Page
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh
Spacing Line