Skip to main content

Automotive clusters

The Automotive initiative uses Red Hat OpenShift Service on AWS (ROSA) for their clusters.

See this internal repo for Infrastructure as Code definitions.

They gave us access to their test cluster - creds are in our internal repo.

There's also a recording of a intro that Hubert Stefanski gave us.

For communication use packit-auto-shared-infra internal chat and/or google group.

Staging packit-service @ test cluster

A new deployment of staging instance of packit-service to the test cluster went flawlessly.

All microservices work OK, the only hiccup we've seen is with PostgreSQL which is working intermittently. The readiness probe sometime fails and/or a query via SQLAlchemy ofter returns "server closed the connection unexpectedly" even when the readiness probe is turned off so the probe can't be the cause per se.

When one runs watch pg_isready in the pod's terminal, the outages are mysteriously gone and don't show up any more even when the watch is later stopped. The readiness probe also calls the pg_isready periodically via /usr/libexec/check-container, but it doesn't have the same effect as running it manually in the terminal.

(author of these letters owe you a beer six pack if you can help shed some light on this mystery)