1060761 - Differences in properties ingested from builds-pending, builds-running & builds-4hr

Status: RESOLVED WONTFIX

(Tree Management :: Treeherder: Data Ingestion, defect, P4)

Description

[Ed Morley [:emorley]]

Following on from bug 1043320, I think there are still a few differences between how we handle builds-pending, builds-running and builds-4hr that may not be intentional (other than the obvious things like "we can't have a start_time if the job is still pending).

For example for pending jobs:
    'machine_platform': {
        'os_name': platform_info['os'],
        'platform': platform_info['os_platform'],
        'architecture': platform_info['arch'],
        'vm': platform_info['vm']
    },

Whereas for completed we're missing "vm":
    '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', '')
    },

Also for completed jobs, we set things like 'product_name' and 'who' (and more), but don't set these for pending, and sometimes not running - even though I'd be surprised if they are not available in builds-{pending,running}.

Probably worth vetting all of these to be sure.

Comment 1

[Ed Morley [:emorley]]

Also ideally we'd remove a lot of the duplication in that file, since there's so much overlap in the handling of builds-pending, builds-running & builds-4hr in buildapi.py.

Comment 2

[Ed Morley [:emorley]]

Wontfix given buildbot is EOL.