Fix resources not sent on service check metrics #14422
Conversation
ghost
added
the
alexandre-normand
changed the title
alexandre-normand
added
changelog/no-changelog
category/bugfix
| _assert_complex_config( | ||
| aggregator, | ||
| [tags.DATABASE_INSTANCE_RESOURCE_TAG.format(hostname='stubbed.hostname')], | ||
| tags.SC_TAGS + [tags.DATABASE_INSTANCE_RESOURCE_TAG.format(hostname='stubbed.hostname')], | ||
| tags.METRIC_TAGS_WITH_RESOURCE, |
There was a problem hiding this comment.
it looks like this is asserting that the dd.resource.database_instance:* would be in this array twice? I am assuming the test case normalizes the tag set somewhere and not that the tags would show up twice?
There was a problem hiding this comment.
Nah, the array only includes the internal resource tag once. tags.SC_TAGS omits the internal tag because it had to be a formatted string depending on the hostname.
There was a problem hiding this comment.
tags.METRIC_TAGS_WITH_RESOURCE includes it, though. Am i just reading this diff incorrectly?
There was a problem hiding this comment.
It does but one of those parameters is the service check tags and the other is the metric tags. They had to be separated because service check tags will include the internal resource tag even in the end to end case while metrics won't have the internal tags (because it gets transformed into actual resources on metrics).
What does this PR do?
While taking over PR 14333, I made the wrong call when addressing the end to end test failure related to the internal resource tags. Instead of removing the resources from service checks that we emit, we actually want to keep the resource tags so that the service check metrics like
check_run.postgres.can_connect.okhave resources set on them.Motivation
Additional Notes
Review checklist (to be filled by reviewers)
changelog/andintegration/labels attachedqa/skip-qalabel.