[DOCS-13132] Expand GCP metric discrepancy guide to cover missing metrics#36309
[DOCS-13132] Expand GCP metric discrepancy guide to cover missing metrics#36309buraizu wants to merge 3 commits into
Conversation
Preview links (active after the
|
|
|
||
| 1. Search for the metric on the [Datadog Metric Summary page][7]. If the metric kind is `CUMULATIVE`, also search for the `.delta` variant (for example, `gcp.gke.container.restart_count.delta`). See [Cumulative metrics][8] for background. | ||
|
|
||
| 2. Confirm the metric is part of the standard Google Cloud Monitoring catalog. The Google Cloud integration does not automatically import [custom metrics][9] or [log-based metrics][10]. |
There was a problem hiding this comment.
I don't think this 100% true; we see Prometheus and user-defined logging metrics get collected in practice. We disable Prometheus metrics by default for new integrations, but we still collect them for many customers. I'd have to do more digging to understand if we collect the other types of custom metrics listed on that page
There was a problem hiding this comment.
Thanks for flagging, I did a bit of digging myself and your concern is legimitate:
- The Incorrect metadata for user defined
gcp.loggingmetrics? section directly contradicts the "does not import log-based metrics" framing. - Google Cloud Managed Service for Prometheus is listed in the supported integrations table on the same page.
Proposing this softer wording for step 2:
Confirm the metric is part of the Google Cloud metric catalog. Some custom metric and log-based metric types are not automatically imported by the integration. If your metric is user-defined or comes from a non-standard source, check the Google Cloud integration page for the current collection behavior, and contact Datadog support if you're unsure.
This drops the absolute claim, keeps the actionable catalog check, and routes uncertainty to support.
There was a problem hiding this comment.
I think we could just say that we don't support Anthos metrics (which are prefixed with kubernetes.io/anthos/ in GCP's naming), but leave out the bit about saying we don't support custom metrics.
2 participants
What does this PR do? What is the motivation?
Fixes DOCS-13132
Expands the Google Cloud metric discrepancy guide (
/integrations/guide/gcp-metric-discrepancy/) to cover both missing metrics and value mismatches. The guide previously covered only alignment-function value mismatches, but escalated GCP support tickets analyzed under DOCS-13132 showed missing-metric confusion was the larger cluster.Changes:
## Metric discrepanciessection with two symptom-scoped sections:## Metric is missing in Datadog(new) and## Metric values do not match the Google Cloud console(existing alignment-functions walkthrough, with original step 1 promoted to a shared section and remaining steps renumbered).## Confirm the metric exists in Google Cloudas a shared first step covering Datadog'sgcp.SERVICE.METRICnaming convention and metric kinds.titleanddescription. Slug is unchanged.The new missing-metric flow walks customers through five self-serve checks: Datadog Metric Summary search (with
.deltareminder forCUMULATIVEmetrics), custom/log-based metric scope, IAM and project access, metric collection filters, and Google Cloud emission verification.Follow-up work is tracked in DOCS-14212: a metric-availability callout on the getting-started page, and SME-led per-service "Not collected" scaffolding for high-traffic GCP service tile pages.
Merge instructions
Merge readiness:
Additional notes