Skip to main content
This page is for teams on an Aspect Enterprise deployment hosted by Aspect, where you ask for configuration changes and Aspect applies them. It covers what to send and what to expect back.

​
Where to send it

Through your Aspect support channel: the shared Slack channel if your agreement includes one, otherwise the support address your Aspect contact gave you. Anything that includes a credential goes through the secure channel, never the chat.

​
What to include

Most of the turnaround on a configuration change is a round trip asking what you meant. These avoid it:
  • Which deployment, if your organization has more than one.
  • What should change, in the terms of the configuration options: the runner group name, the platform, the retention window.
  • What it’s for, in a sentence. Knowing that a new pool is for GPU integration tests rather than general build fan-out often changes what Aspect recommends.
  • When it needs to be live, if there is a date.
For a new runner group or remote execution platform, say what the jobs or actions running there need: architecture, memory, GPU, any system package your actions assume.

​
What’s routine

These are applied without downtime:
  • Scaling a runner group’s minimum or maximum.
  • Adding or removing a runner group.
  • Adding a remote execution platform, or changing a platform’s worker count.
  • Changing which warming set a runner group restores. The targets it warms live in your own aspect ci warming job.
  • Adjusting retention windows.
Scaling changes take effect as the pool next scales. A new runner group needs a matching change in your pipeline to target it, and a new remote execution platform needs the matching Bazel exec platform on your side. Aspect sends the exact platform properties to use, and they have to match verbatim.

​
What needs a window

These are scheduled with you, typically outside your working hours:
  • Version upgrades, including major ones. Aspect proposes a window; see Upgrades.
  • Changing the identity provider, including moving to or from your own identity provider. Every client re-authenticates afterwards.
  • Changing the endpoint domain. Clients must reconfigure: an aspect auth configure against the new host, and a regenerated Bazel rc.
  • Adding private endpoints, which needs coordination on both sides.
  • Storage or database resizing beyond what happens automatically.

​
Upgrades

Aspect schedules and performs them. Patch releases avoid breaking changes and need nothing client-side; a minor or major release can carry breaking changes. Aspect sends any client-side steps before the window, and the carry the detail.

​
Incidents

Aspect monitors the deployment and holds the pager for it. Report an incident that affects you anyway: knowing which of your pipelines is blocked changes the order things get fixed in. If the problem is a Bazel build failing rather than the platform being down, Tuning test resource usage and the Build Results UI are usually faster than a support round trip.