Project is core concept of OneDev, and can be used to hold code, issues, builds, etc. Projects can be organized hierarchically to reflect parent-child relationship as well as facilitate setting inheritance.
If a project does not have parent, its project path will be project name; otherwise, its project path is
<parent project path>/<project name>. Assume below project hierarchy:
Project path of
projectC will be
Git repository, issue attachments, build artifacts and other project related files are placed into project storage directory, which is identified by project id under directory
<OneDev install directory>/site/projects, or
<OneDev data directory>/site/projects for docker installation.
Role defines a set of permissions. When a project is authorized to a user or group, the role must be specified to determine the actual permissions
Like many other git products, OneDev has pull request for work submission and code review. Code review policies can be defined for instance to specify required reviewers when certain files are changed.
Build spec describes specification of build processes in OneDev. It is defined in file
.onedev-buildspec.yml in root of the repository
Jobs are defined in build spec and a single job runs on a single node. Jobs can depend on each other to accomplish complex build workflows such as concurrent cross-node processing. Job can run automatically with job triggers, or manually from desired commits
Job trigger defined when a job should be triggered. Below triggers are currently available:
- Branch update Triggers when new commits are pushed to specified branches
- Tag creation Triggers when specified tags are created
- Pull request open or update Triggers when specified pull requests are opened or updated
- Pull request merge Triggers when specified pull requests are merged
- Pull request discard Triggers when specified pull requests are discarded
- Dependency job finished Triggers when dependency job finishes
- Cron schedule Triggers when scheduled time reaches
Job executors are used to run jobs. There are five types:
- Kubernetes Executor Run jobs as pods in Kubernetes cluster. No agents required
- Server Docker Executor Run jobs inside docker containers on server
- Remote Docker Executor Run jobs inside docker containers on remote machines. Agents need to be installed on remote machines
- Server Shell Executor Run jobs directly with shell/batch facility on server
- Remote Shell Executor Run jobs directly with shell/batch facility on remote machines. Agents need to be installed on remote machines.
Agents runs on remote machines to execute jobs dispatched by server. It can be included in one or more remote job executors via agent selector. Depending on the job executor being used, it executes jobs either inside or outside containers. Agents are not required if you run jobs on server or inside Kubernetes cluster
Job workspace is shared between all steps in same job. It is the working directory of the command step. Repository and dependency files (if there is any) will be retrieved into this directory. Artifacts/reports will also be published from this directory
Steps are defined in job to execute scripts on designated images. Job steps execute sequentially on job node and share the same job workspace
Step template groups a sequence of steps as a whole and can be customized with parameters. A step template can be used in a job and itself can use other step templates if necessary
Services define live facilities used by a job such as database, message queue etc
Build is result of running a job
If some jobs depend on job of a particular build, this build can then be promoted to run these downstream jobs. During promotion, artifacts (normally verified to be
good) of current build can be passed to downstream jobs for further processing
Pipeline is an execution of job dependency graph. Job dependency graph can be executed by running any job in the graph, either manually or automatically via job triggers
Two builds are on same stream if they:
- Belongs to same project
- Belongs to same job
- Corresponding commits are on same branch
When multiple OneDev servers connect to form a cluster (enterprise edition only), lead server is the first started server. You may examine it from menu Administration / High Availability & Scalability
Guest users (since 9.2.1) are limited to read only access to the system, even if they are assigned with more permissions. For instance, if default role of a project grants code write permission, guest users can only get code read permission. They can however open issues and add comments to accessible projects.
More importantly, they are NOT included in user-month calculation in enterprise edition, and are free to use.