Add configsync_sync_generation metric resource tag #763
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
karlkfi
force-pushed
the
karl-gen-label
branch
2 times, most recently
from
ca0b4cb
to
4763360
Compare
|
We could aggregate generation to hide it from GCM, but I'm waiting for #748 to implement that for |
karlkfi
force-pushed
the
karl-gen-label
branch
2 times, most recently
from
7b0d50a
to
58d8e34
Compare
- Add configsync.gke.io/sync-generation reconciler deployment/pod label, populated by the reconciler-manager - Add CONFIGSYNC_SYNC_GENERATION env var on the otel-agent container, populated by the k8s downward API from the pod label - Add configsync.sync.generation metric resource label on the otel-agent config, populated from the env var. This resource label is then converted to a metric label for Prometheus, but not Monarch or Cloud Monitoring, which ignore custom resource labels. - Change the e2e tests to use the sync metric labels, instead of the pod name. This allows the metrics validation to tolerate pod replacement due to rescheduling, oomkill, or autoscaling
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: tiffanny29631 The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
google-oss-prow
bot
merged commit b7ac44e
into
GoogleContainerTools:main
6 checks passed
|
@karlkfi would you update https://cloud.google.com/anthos-config-management/docs/reference/labels-and-annotations to include |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
configsync.gke.io/sync-generationreconciler deployment/pod label, populated by the reconciler-managerCONFIGSYNC_SYNC_GENERATIONenv var on the otel-agent container, populated by the k8s downward API from the pod labelconfigsync.sync.generationmetric resource attribute on the otel-agent config, populated from the env varconfigsync.sync.generationlabel in the otel-collector config, when sending to MonarchThis PR blocks VPA: #691
Motivation
The e2e tests have been using pod name + source commit to filter metrics queries. But pod name was really just a proxy for RSync generation, so that we could match expected metrics with the most recent RSync spec and Git/OCI/Helm source changes. However, it turns out using pod name is still occasionally flaky, because the pod might be replaced or rescheduled even when the RSync spec didn't change (e.g. node replacement, rescheduling, oomkill, or autoscaling). Using RSync generation instead of pod name makes the e2e tests less flaky and makes it possible to add VPA and memory limits, which would otherwise cause most e2e tests to fail metrics validation.
Note:
configsync.sync.generationis a metric resource attribute, NOT a normal metric label. So it will not be sent to either Monarch or Cloud Monitoring, because those providers do not support custom resource attributes without hard-coded support and a custom resource type registered with Google. However, custom resources attributes ARE being sent to Prometheus, because of theresource_to_telemetry_conversion: { enabled: true }setting, which converts them to metric labels. Soconfigsync.sync.generationbecomesconfigsync_sync_generationin Prometheus metrics. This is how the e2e tests use them for query filtering.