Issues workflow

Issue tracker guidelines

Search the issue tracker for similar entries before submitting your own, there’s a good chance somebody else had the same issue or feature proposal. Show your support with an award emoji and/or join the discussion.

Please submit bugs using the ‘Bug’ issue template provided on the issue tracker. The text in the comments (<!-- ... -->) is there to help you with what to include.

Issue triaging

Our issue triage policies are described in our handbook. You are very welcome to help the GitLab team triage issues. We also organize issue bash events once every quarter.

The most important thing is making sure valid issues receive feedback from the development team. Therefore the priority is mentioning developers that can help on those issues. Please select someone with relevant experience from the GitLab team. If there is nobody mentioned with that expertise, look in the commit history for the affected files to find someone.

We also have triage automation in place, described in our handbook.

Labels

To allow for asynchronous issue handling, we use milestones and labels. Leads and product managers handle most of the scheduling into milestones. Labeling is a task for everyone. (For some projects, labels can be set only by GitLab team members and not by community contributors).

Most issues will have labels for at least one of the following:

  • Type. For example: ~"type::feature", ~"type::bug", or ~"type::maintenance".
  • Stage. For example: ~"devops::plan" or ~"devops::create".
  • Group. For example: ~"group::source code", ~"group::knowledge", or ~"group::editor".
  • Category. For example: ~"Category:Code Analytics", ~"Category:DevOps Reports", or ~"Category:Templates".
  • Feature. For example: ~wiki, ~ldap, ~api, ~issues, or ~"merge requests".
  • Department: ~UX, ~Quality
  • Team: ~"Technical Writing", ~Delivery
  • Specialization: ~frontend, ~backend, ~documentation
  • Release Scoping: ~Deliverable, ~Stretch, ~"Next Patch Release"
  • Priority: ~"priority::1", ~"priority::2", ~"priority::3", ~"priority::4"
  • Severity: ~"severity::1", ~"severity::2", ~"severity::3", ~"severity::4"

Please add ~"breaking change" label if the issue can be considered as a breaking change.

Please add ~security label if the issue is related to application security.

All labels, their meaning and priority are defined on the labels page.

If you come across an issue that has none of these, and you’re allowed to set labels, you can always add the type, stage, group, and often the category/feature labels.

Type labels

Type labels are very important. They define what kind of issue this is. Every issue should have one and only one.

The SSOT for type and subtype labels is available in the handbook.

A number of type labels have a priority assigned to them, which automatically makes them float to the top, depending on their importance.

Type labels are always lowercase, and can have any color, besides blue (which is already reserved for category labels).

The descriptions on the labels page explain what falls under each type label.

The GitLab handbook documents when something is a bug and when it is a feature request.

Stage labels

Stage labels specify which stage the issue belongs to.

Naming and color convention

Stage labels respects the devops::<stage_key> naming convention. <stage_key> is the stage key as it is in the single source of truth for stages at https://gitlab.com/gitlab-com/www-gitlab-com/blob/master/data/stages.yml with _ replaced with a space.

For instance, the “Manage” stage is represented by the ~"devops::manage" label in the gitlab-org group since its key under stages is manage.

The current stage labels can be found by searching the labels list for devops::.

These labels are scoped labels and thus are mutually exclusive.

The Stage labels are used to generate the direction pages automatically.

Group labels

Group labels specify which groups the issue belongs to.

It’s highly recommended to add a group label, as it’s used by our triage automation to infer the correct stage label.

Naming and color convention

Group labels respects the group::<group_key> naming convention and their color is #A8D695. <group_key> is the group key as it is in the single source of truth for groups at https://gitlab.com/gitlab-com/www-gitlab-com/blob/master/data/stages.yml, with _ replaced with a space.

For instance, the “Pipeline Execution” group is represented by the ~"group::pipeline execution" label in the gitlab-org group since its key under stages.manage.groups is pipeline_execution.

The current group labels can be found by searching the labels list for group::.

These labels are scoped labels and thus are mutually exclusive.

You can find the groups listed in the Product Stages, Groups, and Categories page.

We use the term group to map down product requirements from our product stages. As a team needs some way to collect the work their members are planning to be assigned to, we use the ~group:: labels to do so.

Category labels

From the handbook’s Product stages, groups, and categories page:

Categories are high-level capabilities that may be a standalone product at another company, such as Portfolio Management, for example.

It’s highly recommended to add a category label, as it’s used by our triage automation to infer the correct group and stage labels.

If you are an expert in a particular area, it makes it easier to find issues to work on. You can also subscribe to those labels to receive an email each time an issue is labeled with a category label corresponding to your expertise.

Naming and color convention

Category labels respects the Category:<Category Name> naming convention and their color is #428BCA. <Category Name> is the category name as it is in the single source of truth for categories at https://gitlab.com/gitlab-com/www-gitlab-com/blob/master/data/categories.yml.

For instance, the “DevOps Reports” category is represented by the ~"Category:DevOps Reports" label in the gitlab-org group since its devops_reports.name value is “DevOps Reports”.

If a category’s label doesn’t respect this naming convention, it should be specified with the label attribute in https://gitlab.com/gitlab-com/www-gitlab-com/blob/master/data/categories.yml.

Feature labels

From the handbook’s Product stages, groups, and categories page:

Features: Small, discrete functionalities, for example Issue weights. Some common features are listed within parentheses to facilitate finding responsible PMs by keyword.

It’s highly recommended to add a feature label if no category label applies, as it’s used by our triage automation to infer the correct group and stage labels.

If you are an expert in a particular area, it makes it easier to find issues to work on. You can also subscribe to those labels to receive an email each time an issue is labeled with a feature label corresponding to your expertise.

Examples of feature labels are ~wiki, ~ldap, ~api, ~issues, and ~"merge requests".

Naming and color convention

Feature labels are all-lowercase.

Facet labels

To track additional information or context about created issues, developers may add facet labels. Facet labels are also sometimes used for issue prioritization or for measurements (such as time to close). An example of a facet label is the ~customer label, which indicates customer interest.

Department labels

The current department labels are:

  • ~UX
  • ~Quality

Team labels

Important: Most of the historical team labels (like Manage or Plan) are now deprecated in favor of Group labels and Stage labels.

Team labels specify what team is responsible for this issue. Assigning a team label makes sure issues get the attention of the appropriate people.

The current team labels are:

  • ~Delivery
  • ~"Technical Writing"

Naming and color convention

Team labels are always capitalized so that they show up as the first label for any issue.

Specialization labels

These labels narrow the specialization on a unit of work.

  • ~frontend
  • ~backend
  • ~documentation

Release scoping labels

Release Scoping labels help us clearly communicate expectations of the work for the release. There are three levels of Release Scoping labels:

  • ~Deliverable: Issues that are expected to be delivered in the current milestone.
  • ~Stretch: Issues that are a stretch goal for delivering in the current milestone. If these issues are not done in the current release, they will strongly be considered for the next release.
  • ~"Next Patch Release": Issues to put in the next patch release. Work on these first, and add the ~"Pick into X.Y" label to the merge request, along with the appropriate milestone.

Each issue scheduled for the current milestone should be labeled ~Deliverable or ~"Stretch". Any open issue for a previous milestone should be labeled ~"Next Patch Release", or otherwise rescheduled to a different milestone.

Priority labels

We have the following priority labels:

  • ~"priority::1"
  • ~"priority::2"
  • ~"priority::3"
  • ~"priority::4"

Please refer to the issue triage priority label section in our handbook to see how it’s used.

Severity labels

We have the following severity labels:

  • ~"severity::1"
  • ~"severity::2"
  • ~"severity::3"
  • ~"severity::4"

Please refer to the issue triage severity label section in our handbook to see how it’s used.

Label for community contributors

There are many issues that have a clear solution with uncontroversial benefit to GitLab users. However, GitLab might not have the capacity for all these proposals in the current roadmap. These issues are labeled ~"Seeking community contributions" because we welcome merge requests to resolve them.

Community contributors can submit merge requests for any issue they want, but the ~"Seeking community contributions" label has a special meaning. It points to changes that:

  1. We already agreed on,
  2. Are well-defined,
  3. Are likely to get accepted by a maintainer.

We want to avoid a situation when a contributor picks an ~"Seeking community contributions" issue and then their merge request gets closed, because we realize that it does not fit our vision, or we want to solve it in a different way.

We manually add the ~"Seeking community contributions" label to issues that fit the criteria described above. We do not automatically add this label, because it requires human evaluation.

We recommend people that have never contributed to any open source project to look for issues labeled ~"Seeking community contributions" with a weight of 1 or the ~"good for new contributors" label attached to it. More experienced contributors are very welcome to tackle any of them.

For more complex features that have a weight of 2 or more and clear scope, we recommend looking at issues with the label ~"Community Challenge". If your MR for the ~"Community Challenge" issue gets merged, you will also have a chance to win a custom GitLab merchandise.

If you’ve decided that you would like to work on an issue, please @-mention the appropriate product manager as soon as possible. The product manager will then pull in appropriate GitLab team members to further discuss scope, design, and technical considerations. This will ensure that your contribution is aligned with the GitLab product and minimize any rework and delay in getting it merged into main.

GitLab team members who apply the ~"Seeking community contributions" label to an issue should update the issue description with a responsible product manager, inviting any potential community contributor to @-mention per above.

Stewardship label

For issues related to the open source stewardship of GitLab, there is the ~"stewardship" label.

This label is to be used for issues in which the stewardship of GitLab is a topic of discussion. For instance if GitLab Inc. is planning to add features from GitLab EE to GitLab CE, related issues would be labeled with ~"stewardship".

A recent example of this was the issue for bringing the time tracking API to GitLab CE.

Feature proposals

To create a feature proposal, open an issue on the issue tracker.

In order to help track the feature proposals, we have created a ~"type::feature" label. For the time being, users that are not members of the project cannot add labels. You can instead ask one of the core team members to add the label ~"type::feature" to the issue or add the following code snippet right after your description in a new line: ~"type::feature".

Please keep feature proposals as small and simple as possible, complex ones might be edited to make them small and simple.

Please submit feature proposals using the ‘Feature Proposal’ issue template provided on the issue tracker.

For changes to the user interface (UI), follow our design and UI guidelines, and include a visual example (screenshot, wireframe, or mockup). Such issues should be given the ~UX" label for the Product Design team to provide input and guidance. You may need to ask one of the core team members to add the label, if you do not have permissions to do it by yourself.

If you want to create something yourself, consider opening an issue first to discuss whether it is interesting to include this in GitLab.

Issue weight

Issue weight allows us to get an idea of the amount of work required to solve one or multiple issues. This makes it possible to schedule work more accurately.

You are encouraged to set the weight of any issue. Following the guidelines below will make it easy to manage this, without unnecessary overhead.

  1. Set weight for any issue at the earliest possible convenience
  2. If you don’t agree with a set weight, discuss with other developers until consensus is reached about the weight
  3. Issue weights are an abstract measurement of complexity of the issue. Do not relate issue weight directly to time. This is called anchoring and something you want to avoid.
  4. Something that has a weight of 1 (or no weight) is really small and simple. Something that is 9 is rewriting a large fundamental part of GitLab, which might lead to many hard problems to solve. Changing some text in GitLab is probably 1, adding a new Git Hook maybe 4 or 5, big features 7-9.
  5. If something is very large, it should probably be split up in multiple issues or chunks. You can not set the weight of a parent issue and set weights to children issues.

Regression issues

Every monthly release has a corresponding issue on the CE issue tracker to keep track of functionality broken by that release and any fixes that need to be included in a patch release (see 8.3 Regressions as an example).

As outlined in the issue description, the intended workflow is to post one note with a reference to an issue describing the regression, and then to update that note with a reference to the merge request that fixes it as it becomes available.

If you’re a contributor who doesn’t have the required permissions to update other users’ notes, please post a new note with a reference to both the issue and the merge request.

The release manager will update the notes in the regression issue as fixes are addressed.

Technical and UX debt

In order to track things that can be improved in the GitLab codebase, we use the ~"technical debt" label in the GitLab issue tracker. We use the ~"UX debt" label when we choose to deviate from the MVC, in a way that harms the user experience.

These labels should be added to issues that describe things that can be improved, shortcuts that have been taken, features that need additional attention, and all other things that have been left behind due to high velocity of development. For example, code that needs refactoring should use the ~"technical debt" label, something that didn’t ship according to our Design System guidelines should use the ~"UX debt" label.

Everyone can create an issue, though you may need to ask for adding a specific label, if you do not have permissions to do it by yourself. Additional labels can be combined with these labels, to make it easier to schedule the improvements for a release.

Issues tagged with these labels have the same priority like issues that describe a new feature to be introduced in GitLab, and should be scheduled for a release by the appropriate person.

Make sure to mention the merge request that the ~"technical debt" issue or ~"UX debt" issue is associated with in the description of the issue.

Technical debt in follow-up issues

It’s common to discover technical debt during development of a new feature. In the spirit of “minimum viable change”, resolution is often deferred to a follow-up issue. However, this cannot be used as an excuse to merge poor-quality code that would otherwise not pass review, or to overlook trivial matters that don’t deserve to be scheduled independently, and would be best resolved in the original merge request - or not tracked at all!

The overheads of scheduling, and rate of change in the GitLab codebase, mean that the cost of a trivial technical debt issue can quickly exceed the value of tracking it. This generally means we should resolve these in the original merge request - or not create a follow-up issue at all.

For example, a typo in a comment that is being copied between files is worth fixing in the same MR, but not worth creating a follow-up issue for. Renaming a method that is used in many places to make its intent slightly clearer may be worth fixing, but it should not happen in the same MR, and is generally not worth the overhead of having an issue of its own. These issues would invariably be labeled ~P4 ~S4 if we were to create them.

More severe technical debt can have implications for development velocity. If it isn’t addressed in a timely manner, the codebase becomes needlessly difficult to change, new features become difficult to add, and regressions abound.

Discoveries of this kind of technical debt should be treated seriously, and while resolution in a follow-up issue may be appropriate, maintainers should generally obtain a scheduling commitment from the author of the original MR, or the engineering or product manager for the relevant area. This may take the form of appropriate Priority / Severity labels on the issue, or an explicit milestone and assignee.

The maintainer must always agree before an outstanding discussion is resolved in this manner, and will be the one to create the issue. The title and description should be of the same quality as those created in the usual manner - in particular, the issue title must not begin with Follow-up! The creating maintainer should also expect to be involved in some capacity when work begins on the follow-up issue.