Architecture Data Flow
⚠️ Note!You’ll need access to the appropriate groups to work with disruption data, please reach out to the TRT team for access.
DPCR Job Aggregation Configs(private repo)
Disruption Data Architecture
BigQueryto generate the disruption historical data. It does not cover how the tests themselves are run against a cluster.
High Level Diagram
How The Data Flows
disruption samplers, these run
GETrequests against a number of backends in the cluster every second and record the results to determine disruption. (see Testing Backends For Availability for more info)
Disruption Uploaderis a
CronJobthat is set to run every
4 hours. All the
Uploaderjobs are run in the DPCR cluster, the current configuration can be found in the
openshift/continuous-release-jobsprivate repo under
When e2e tests are done the results are uploaded to
GCSand the results can be viewed in the artifacts folder for a particular job run.
Clicking the artifact link on the top right of a prow job and navigating to the
openshift-e2e-testfolder will show you the disruption results. (ex.
We only pull disruption data for job names specified in the
BigQuery. (see Job Primer for more information on this process)
The disruption uploader will parse out the results from the e2e run backend-disruption json files and push them to the openshift-ci-data-analysis table in BigQuery.
Currently, backend disruption data is queried from BigQuery and downloaded in
jsonformat. The resulting
jsonfile is then committed to origin/pkg/synthetictests/allowedbackenddisruption/query_results.json for backend disruption or origin/pkg/synthetictests/allowedalerts/query_results.json for alert data (see how to query)
openshift/originare then used by the the matchers that the
samplersinvoke to find the best match for a given test (typically with “remains available using new/reused connections” or “should be nearly zero single second disruptions”) to check if we’re seeing noticeably worse disruption during the run.
How To Query The Data
Once you have access to BigQuery in the
openshift-ci-data-analysis project, you can run the below query to fetch the latest results.
Once the query is run, you can download the data locally.