redhat-performance / cloud-governance / 30354115328
73%

DEFAULT BRANCH: main
Ran
Jobs 0
Files 0
Run time –
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
30354115328

push

travis-pro

web-flow 
Add Orion metrics rollup job (#1017)

* Add Orion metrics rollup job

Rolls up per-resource policy execution data from
cloud-governance-policy-es-index into one document per account per
day, tracking five policies (zombie_cluster_resource, s3_inactive,
ec2_stop, unused_access_key, delete_access_key) plus an aggregate
monitored_policies_savings figure, into a new dedicated index. This
is the shape validated against Orion (change-point regression
detection) in an earlier spike.

Wired in via the existing cost_policies dispatch mechanism (same
pattern as yearly_savings_report) rather than touching main.py
directly. Reads from a hardcoded source index rather than the shared
es_index env var, which defaults to the wrong index for any
cost_policies member. Supports both a one-time backfill (chunked by
month, mirroring yearly_savings_report's pattern) and the default
daily incremental mode.

* Broaden monitored_policies_savings to all policies, not just tracked ones

Confirmed against production data: none of the 5 tracked policies
(zombie_cluster_resource, s3_inactive, ec2_stop, unused_access_key,
delete_access_key) ever populate TotalYearlySavings - they're
risk/hygiene policies, not idle-resource-elimination policies like
ip_unattached/unused_nat_gateway that do. Scoping the savings sum to
just the tracked policies made it permanently 0, giving Orion nothing
to detect. Removed the top-level policy filter so the savings sum
sees every policy; per-policy resource counts stay scoped to the
tracked set via 'include' on the nested terms aggregation instead.

Verified live against the real OpenSearch cluster: monitored_policies_savings
now returns real, varying, non-zero values.

* Warn on partial backfill date range instead of silently ignoring it

If only one of start_date/end_date resolves truthy, the job was
silently falling back to daily mode with no indication the provided
date was ignored - a misconfigured backfill (e.g. start date set, end
date forgotte... (continued)
Source Files on build 30354115328
Detailed source file information is not available for this build.