Mountain View 2016

Accounts in NoteDb

NoteDb is our effort to migrate the Gerrit meta data from the database into the Git repositories. This migration also includes the Gerrit accounts. This talk will discuss how the accounts will be stored in Git.

Slides (PDF)

Edwin Kempin, Google

Atomicity with change-sets and topic submission

In our software development environment, with source code spread across hundreds of source code repositories, related changes created in a same source code repository and or across multiple source code repositories need to be reviewed, approved, tested and committed together. Developers creating changes across multiple repositories need ability to group the changes together as a unit. The grouping of changes should be intact across all re-write operations in GIT like commit amend and rebase. Developers need the flexibility to group the related changes implicitly or explicitly. Developers need the ability to work on multiple related change-sets simultaneously in the same sandbox. While developers need ability to group the changes flexibly, the system needs the ability to detect the dependencies between the change-sets created in the same sandbox .

Come and find out how we use topic submission feature of Gerrit along with wrapper scripts, hooks and plugin to provide atomicity for changes throughout lifecycle of related changes.

Slides (PDF)

Basavaraj Kradakal, Juniper Networks

Implementing Gerrit build with Bazel

Gerrit build tool chain is an interesting journey with multiple stops:

  • Maven
  • Buck
  • Blaze, opps.. Bazel


David Ostrovsky, Unaffiliated

Bringing pull requests to Gerrit

With CollabNet TeamForge it is now possible to use feature branch workflow in addition to standard gerrit workflow to work on your changes. In this presentation you will learn how it works, why we have decided to implement it, how was it implemented and what were the choices we have made and challenges along the way.

Slides (slideshare)

Eryk Szymanski, CollabNet

Plugins for CI Systems

I will give an overview of 3 new plugins developed by the Qualcomm Innovation Center for use by CI systems: the batch, manifest, and task plugins. I will review what these plugins can do, and how we use them along with our future plans and ideas for these plugins. I will also talk about the core modifications that were needed to create these plugins, and what future core modification I believe may still be needed, along with some of the challenges of doing more in plugins.

A quick overview of the 3 plugins:

The batch plugin provides a mechanism for building and previewing sets of proposed updates to multiple projects/branches/refs that should be applied together. The focus of batch updates tend to be verification (by CI systems). The batch update service provides the tools to build refs by merging changes to temporary “snapshot” refs, which can then be tested extensively, and finally submitted ”as is".

The manifest plugin provides server side utilities to operate on, and query information about repo manifests (xml) stored in git projects on the current server. This plugin provides APIs to update values in manifests, and to search for manifests with certain values.

The task plugin provides a mechanism to manage tasks which need to be performed on changes along with a way to expose and query this information. Tasks are organized hierarchically, and task definitions use gerrit queries to define which changes each task applies to, and how to define the status criteria for each task. An important use case of the task plugin is to have a common place for CI systems to define which changes they will operate on, and when they will do so.

Martin Fick, Qualcomm Innovation Center

Email Ingestion

Email ingestion gives users the option to reply to auto-generated email notifications from Gerrit, have their comments parsed out from the email and attached to the change. Email ingestion is currently under development and this session will be about the design, implementation and usage of the new feature.

Patrick Hiesel, Google

So what's up with the repo tool and submodules?

This talk will present the current situation for submodules in Git, Gerrits ability to deal with the awesomeness of submodules and how we keep git-repo around.

Slides (PDF)

Stefan Beller, Google

Gerrit Analytics

Gerrit Review Analytics Dashboards. Collect, analyze and display in Real Time the Gerrit Code Review activity to create GitHub-style dashboards and archive the review data audit trail for off-line reporting.


Luca Milanesio, GerritForge

Building Gerrit - a CI story

It is one year since the official introduction of a CI build for Gerrit. Learn how we implemented a fully featured build pipeline for Gerrit and its plugins, ranging from Maven, Simple Build Tool, Buck and more recently Bazel-powered build scripts.

Luca Milanesio, Maintainer of Gerrit-CI

Introducing ‘zoekt’: git aware codesearch

Zoekt is an open-source, fast full-text search engine for code, made to work well for Git repositories.

Slides (PDF)

Han-Wen Nienhuys, Google

Infinite Gerrit

Gerrit Review with Apache Cassandra as JGit storage.

Achieve a “virtually infinite” extensibility of the Gerrit repositories by using a multi-site distributed storage backend based on Apache Cassandra.

Benefits are:

  1. Steps towards a real Gerrit peer-to-peer infrastructure where every server can potentially read and write to any of the Cassandra nodes.
  2. Ability to extend the capacity of the existing network of Gerrit servers by adding additional Cassandra nodes and getting the quorum distribution out-of-the-box.

Slides (slideshare)

Haithem Jarraya, GerritForge

Keep the conversation flowing

A typical Gerrit installation contains integration with an automated testing system that evaluates patchsets and reports results to Gerrit. The only way for a Continous Integration system to report results to Gerrit is by posting a review as a comment. The problem with this workflow is that automated reviews and human reviews are stored as one piece of data (comments). Human reviews are inherently different than automated reviews. Human reviews have more meaning to other human reviewers, it serves as a conversation between people that are reviewing the change and thus it is typically given higher priority over automated reviews. Comments provide a great forum to discuss a change however when robots clutter the forum it overwhelms human reviewers and thus impede the discussion. Robots should have a separate feedback channel so that the data can be easily queried, viewed and analyzed independently from human comments.

This is where the verify-status plugin may help. It creates a separate “verify-status” channel for automated system to report test results. It provides a set of SSH commands and REST endpoints allowing easy integration with any CI system. It allows the verify-status data to be stored in the Gerrit database or on a completely separate database. It provides a set of UI components to view the data independent of Gerrit comments. Lastly there's even a Jenkins verify-status-reporter plugin that helps Jenkins report results to gerrit using this new communications channel.

This talk with go over the motivation behind this plugin, it's status, how we plan to use it, and how it can help you keep the conversation flowing.

Slides (PDF)

Khai Do, OpenStack

What's New in Gerrit 2.12, 2.13 and 2.14

Overview of the new features included in the latest Gerrit releases 2.12 and 2.13, and a preview of what's coming in 2.14.

Slides (PDF)

David Pursehouse, CollabNet

An Update on PolyGerrit

PolyGerrit has been under development for more than a year and is rapidly approaching feature parity with the current GWT UI. This talk will give an overview of what’s been done, what’s left, and what’s next.

Come hear what we’ve learned from a product development standpoint and how you can help us further improve the best code review experience out there.

Slides (PDF)

Andrew Bonventre, Google

Robot Comments

We are extending Gerrit to better support comments generated by automated systems (“robot comments”). Robot comments will be specifically visualized in the UI and users will be able to apply filters on them. Robot comments can also carry generated fixes that users can apply to the change and create a new patch set.

This talk gives you an introduction to robot comments and how they will be implemented in Gerrit.

Edwin Kempin, Google

Slides (PDF)

Zero-downtime Gerrit upgrades

How to upgrade a Gerrit production instance without pulling the plug. Even without having a fully redundant multi-master setup like Google it is possible to perform a full Gerrit upgrade without having to interrupt incoming traffic.

Learn how managed to:

  • upgrade its DB
  • migrate data-center from Europe to America
  • update Gerrit binaries and plugins
  • performed full reindex

... and keeping its 100% availability at the same time.


Luca Milanesio,GerritForge