You are here

Further AFS development and automation

Project ID: 
135
Current stage: 
Manager: 
What: 

Description: Although AFS is considerably easier to manage than NFS, there is still a considerable amount of day to day work to be done in managing the file system, much of which could be automated, This project will investigate ways in which the management workload of AFS could be reduced and produce the appropriate tools.

Deliverables: A suite of scripts and tools reducing the management workload of the AFS file system. Examples are wrapper scripts for long running jobs, an easier way of managing ACLS and a mountpoint database (carried over from the original AFS project)

Why: 

Customer: School

Case statement: One of the School's strategic aims is to reduce the cost of managing core services through automation of operational tasks. This project will contribute to that aim.

When: 

Status:

Timescales:

Priority: Pretty high.

Time: The project is expected to take about five weeks of effort.

How: 

Proposal:

Resources: Two or three virtual machines.

Plan: The plan behind the plan is to fulfil as many as possible of the needs listed in the Services Unit AFS Enhancements page. The plan is to build an OpenAFS installation with its own test cell, distinct from the School's AFS cell, and to use that in tackling the enhancements in order of priority (as marked on the page).

  1. Automate conversion of RO volumes to RW
    1. Establish a test cell
    2. Create some RW and RO volumes on separate servers
    3. Play about with server disasters and with converting RO volumes to RW, check that I have The Right Way To Do It
    4. Play about with finding RW volumes and their RO volumes on other servers
    5. Script the safe conversion of one RO volume to RW
    6. Script the safe replacement of a defunct RW volume via the conversion of a RO mirror elsewhere to RW
    7. Also remove from the VLDB any RO mirrors on the same partition as the original RW volume
    8. Also add a RO volume on the same partition as the new RW volume
    9. Expand the script to also deal with an entire partition's RO volumes
    10. Expand the script again to deal with every partition on a server.
  2. (Expand plan for other priorities)
Other: 

Dependencies: None

Risks:

Milestones

Proposed date Achieved date Name Description