[LU-2064] HSM - Changes to Test Framework - Landing to Master

• Type: Story
• Resolution: Fixed
• Priority: Critical
• Fix Version/s: Lustre 2.5.0
• Affects Version/s: Lustre 2.5.0
• Labels: HSM
• Project: HSM
• Assignee: Jodi Levi (Inactive)
• Reporter: Jodi Levi (Inactive)

Description

This is for patches landing to master

Issue Links

• is related to: LU-3608 HSM Master Ticket for 2.5 Landings

Comments

[Chris Gearing (Inactive)]

The steps moving forwards will be for Yu Jian and I to discuss implementation whilst ensuring CEA are completely in the loop about the details. We have a meeting planned for the week starting 8th July 2013.

[Chris Gearing (Inactive)]

This document describes at a high level the changes required to support HSM. https://jira.hpdd.intel.com/secure/attachment/11767/HSM-16.pdf

I think this can be broken down into two parts; Changes to autotest and changes to the framework.

The link between them that will be in the config file will look like this;

CPTCOUNT=2
CPT1_HOST=node_23
CPT2_HOST=node_24
CPT1DEV0=/dev/sdr1
CPT2DEV0=/dev/sde3

Note; I have changed from AGT (Agent which is a generic term) to CPT (copytool which is more specific)

I'm not sure what AGTSINGLE means so have not replicated it.

I also prepose the CPTxDEVy format so we can have multiple copy tools with multiple devices if need be. Although this might cause problems in autotest/lustre because the presumption is that for all OSSs for example (the devices are the same so they have OSTDEV1, OSTDEV2 which must be true for all OSS's. i.e. OSS must be identical for test.

I will ask Jacques-Charles for the source of the document so we can edit it.