Closed Bug 1056928 Opened 10 years ago Closed 2 years ago

Fix machine_platform or else remove it if it's not needed

Categories

(Tree Management :: Treeherder: Data Ingestion, defect, P3)

defect

Tracking

(Not tracked)

RESOLVED WONTFIX

People

(Reporter: emorley, Unassigned)

Details

iirc we discussed this on the whiteboard at the first work week, but to refresh: The OS/platform/architecture values for a job can actually be referring to several things... 1) The os/platform/arch of the machine that ran the compilation of the build 2) The os/platform/arch of the target device that will run the build 3) The os/platform/arch of the machine that ran the tests Example: An Android build is compiled on a Linux64 machine (#1), targeted at Android ARMv7 (#2), and then tests get run on it on Android 4.0 Pandaboards (#3). For a build job of this type, I would expect a job arch/platform of "armv7"/"Android" and a machine arch/platform of "x86_64"/"Linux". For the test job run on a pandaboard, I would expect a job arch/platform of "armv7"/"Android 4.0", an identical machine arch/platform of "armv7"/"Android 4.0". Whilst doing bug 1040418 I noticed that we do: platform_info = buildbot.extract_platform_info(prop['buildername']) ... #build_platform same as machine_platform 'build_platform': { #platform attributes sometimes parse without results 'os_name': platform_info.get('os', ''), 'platform': platform_info.get('os_platform', ''), 'architecture': platform_info.get('arch', '') }, 'machine_platform': { 'os_name': platform_info.get('os', ''), 'platform': platform_info.get('os_platform', ''), 'architecture': platform_info.get('arch', '') }, Which is setting machine_platform to the same values as build_platform.
Is machine_platform used for anything at present? If it's not surfaced in the UI and isn't affecting the handling of results, this doesn't have to block.
Flags: needinfo?(mdoglio)
The UI shows both machine and build platform in the job detail panel and the job results are aggregated in the service based on a combination of machine platform and build option. We can very quickly change that to be the build platform if we need to, the choice there was pretty random I suppose.
Flags: needinfo?(mdoglio)
From vidyo: since the values are identical, it doesn't matter we are aggregating on machine_platform for now (so this isn't a blocker), however we should switch over to using the job platform everywhere at some point.
Priority: P1 → P3
Component: Treeherder → Treeherder: Data Ingestion
Priority: P3 → P4
Priority: P4 → P3
Summary: machine_platform is set to the values of the job not the machine → Fix machine_platform or else remove it if it's not needed

cleaning out old bugs

Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.