Arvados

Note: for this ticket examples the assumption is that there are 600 blocks equally distributed in the volumes.

• to have all the expected behaviour for keep-balance
• to add to the documentation this behaviour
• to add necesary tests to make sure we comply with this behaviour

Volumes scenario¶

Scenario 1

---
Clusters:
  test1:
    Collections:
      DefaultReplication: 1
    Volumes:
      test1-nyw5e-000000000000000:
        AccessViaHosts:
          'http://keep0:25107': {}
      test1-nyw5e-111111111111111:
        AccessViaHosts:
          'http://keep1:25108': {}

Scenario 2

---
Clusters:
  test1:
    Collections:
      DefaultReplication: 1
    Volumes:
      test1-nyw5e-000000000000000:
        AccessViaHosts:
          'http://keep0:25107': {}
      test1-nyw5e-111111111111111:
        AccessViaHosts:
          'http://keep1:25108': {}
        StorageClasses:
          archival: true

Scenario 3

---
Clusters:
  test1:
    Collections:
      DefaultReplication: 1
    Volumes:
      test1-nyw5e-000000000000000:
        AccessViaHosts:
          'http://keep0:25107': {}
      test1-nyw5e-111111111111111:
        AccessViaHosts:
          'http://keep1:25108': {}
        StorageClasses:
          archival: true
        Read-Only

Expected result after tunnign keep balance¶

To be talked in next grooming session (Feb 16th)

• What are the common usecase ?
• What other scenarios we care about?
• How we address the process of changing configuration and re-running keep-balance?