CI pipelines in prod one fail build during test Intelligence step

Incident Report for Harness

Postmortem

Impact:

In Prod1 environment, the tests page was not responsive on some CI pipelines. The page loads intermittently.

Root Cause:

The TI API handlers talk to TimeScale DB to fetch the test summary data which in turn hook into a distributed tracing framework. The framework SDK calls were taking longer than expected resulting in the Test Report APIs (upload and fetch) timing out.

Timeline:

11:54 AM PDT Internal FireHydrant got created.

12:08 PM PDT Deployed the TI service with tracing disabled.

Downtime: 15 mins after FireHydrant was triggered. Partial (only CI pipelines with test reports configured)

Action Items:

  • Remove tracing for TI service which has been known to cause instability.
  • Enhance logging
  • Add pprof API endpoints to the TI service to debug issues in QA and pre-QA.
Posted Oct 12, 2023 - 06:21 PDT

Resolved

We can confirm normal operation. Get Ship Done!
We will continue to continue to monitor and ensure stability.
Posted Mar 29, 2023 - 00:08 PDT

Investigating

Harness has identified failures of pipelines during the build phase. This is a placeholder message.
Posted Mar 28, 2023 - 23:54 PDT
This incident affected: Prod 1 (Continuous Integration Enterprise(CIE) - Mac Cloud Builds).