From Fedora Project Wiki
Line 20: Line 20:
== Benefit to Fedora ==
== Benefit to Fedora ==


The benefit to Release Engineering should be self-evident: less tools, hosts, and services to maintain. This opens the door a little to let a small set of users generate their own signed repositories from Koji if they want to as well. This may be useful for QA testing.
The benefit to Release Engineering should be self-evident: less tools, hosts, and services to maintain. This opens the door a little to let a small set of users generate their own signed repositories from Koji if they want to as well. This may be useful for QA testing. Bodhi may also make use of this feature to release updates to users.
 
 
== Scope ==
== Scope ==
* Proposal owners: Jay Greguske
* Proposal owners: Jay Greguske

Revision as of 21:20, 9 November 2015

Koji Generates Repositories of Signed RPMs

Summary

Extend Koji with a new feature that allows users to generate yum repositories of signed RPMs.

Owner

  • Name: Jay Greguske
  • Email: jgregusk with the usual red hat domain
  • Release notes owner:

Current status

  • Targeted release: Fedora 24
  • Last updated: 2015-11-09
  • Tracker bug: <will be assigned by the Wrangler>

Detailed Description

This is a significant enabler for generating DVD media, other ISOs, and images more efficiently. It also allows other tools such as mash or pungi to offload much of the heavy-lifting to the build system. Longer term, we may be able to reduce the number of tools needed to manufacture Fedora releases.

Benefit to Fedora

The benefit to Release Engineering should be self-evident: less tools, hosts, and services to maintain. This opens the door a little to let a small set of users generate their own signed repositories from Koji if they want to as well. This may be useful for QA testing. Bodhi may also make use of this feature to release updates to users.

Scope

  • Proposal owners: Jay Greguske
  • Release engineering: This feature does require coordination with release engineering (e.g. changes to installer image generation or update package delivery.)

Upgrade/compatibility impact

No effect on Fedora users directly.

In the unlikely event that Fedora Rel-Eng needs to restore a release from scratch (from a compromised host, or major data loss event) the old process will still need to exist for Fedora 23.

How To Test

The feature affects how Fedora is composed, how images are generated, and potentially how the distribution is delivered. It does not affect the software itself, so this does not bring in any additional functional testing. Composes generated with the new Koji feature should function just as the previous ones did with no disruption to QA consumption of the content.

This does not require any new hardware nor actions from the user base.

User Experience

This feature should not affect user experience in any way.

Dependencies

This work does not depend on any external factors within the distro.

Pungi and mash's future implementations do have a dependency on this work. This is an enabler to other future Koji features as well.

Contingency Plan

  • Contingency mechanism: Just use the old process for generating Fedora composes and delivering them.
  • Contingency deadline: N/A (not a System Wide Change)
  • Blocks release? Probably not.
  • Blocks product? Nope

Documentation

This should not affect Fedora documentation for users, but it will affect Standard Operating Procedures (SOPs) for Rel-Eng. If the current process to build and release Fedora is documented anywhere, it will need to be updated.

Release Notes