Documentation
¶
Index ¶
- func DebugHandler() http.HandlerFunc
- func FilterPipelineRunByState(ctx context.Context, tekton versioned2.Interface, orderList []string, ...) []string
- func PrKey(run *tektonv1.PipelineRun) string
- func RegisterForDebug(qm *Manager)
- func RepoKey(repo *v1alpha1.Repository) string
- type Manager
- func (qm *Manager) AddListToRunningQueue(repo *v1alpha1.Repository, list []string) ([]string, error)
- func (qm *Manager) AddToPendingQueue(repo *v1alpha1.Repository, list []string) error
- func (qm *Manager) InitQueues(ctx context.Context, tekton versioned2.Interface, pac versioned.Interface) error
- func (qm *Manager) QueuedPipelineRuns(repo *v1alpha1.Repository) []string
- func (qm *Manager) RemoveAndTakeItemFromQueue(repo *v1alpha1.Repository, run *tektonv1.PipelineRun) string
- func (qm *Manager) RemoveFromQueue(repoKey, prKey string) bool
- func (qm *Manager) RemoveRepository(repo *v1alpha1.Repository)
- func (qm *Manager) RunningPipelineRuns(repo *v1alpha1.Repository) []string
- func (qm *Manager) Snapshot() (map[string]RepoQueue, bool)
- type ManagerInterface
- type RepoQueue
- type Semaphore
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func DebugHandler ¶ added in v0.50.0
func DebugHandler() http.HandlerFunc
DebugHandler serves a read-only view of the queues as JSON. It mutates nothing and exposes only PipelineRun names, which the caller can already list from the cluster.
func FilterPipelineRunByState ¶
func FilterPipelineRunByState(ctx context.Context, tekton versioned2.Interface, orderList []string, wantedStatus, wantedState string) []string
FilterPipelineRunByInProgress filters the given list of PipelineRun names to only include those that are in a "queued" state and have a pending status. It retrieves the PipelineRun objects from the Tekton API and checks their annotations and status to determine if they should be included.
Returns A list of PipelineRun names that are in a "queued" state and have a pending status.
func PrKey ¶
func PrKey(run *tektonv1.PipelineRun) string
func RegisterForDebug ¶ added in v0.50.0
func RegisterForDebug(qm *Manager)
RegisterForDebug makes a Manager reachable from DebugHandler. The Manager is built long after the watcher's HTTP server is set up, so the handler has to find it later rather than be handed it up front.
func RepoKey ¶
func RepoKey(repo *v1alpha1.Repository) string
Types ¶
type Manager ¶
type Manager struct {
// contains filtered or unexported fields
}
func NewManager ¶
func NewManager(logger *zap.SugaredLogger) *Manager
func (*Manager) AddListToRunningQueue ¶
func (qm *Manager) AddListToRunningQueue(repo *v1alpha1.Repository, list []string) ([]string, error)
AddListToRunningQueue adds the pipelineRun to the waiting queue of the repository and if it is at the top and ready to run which means currently running pipelineRun < limit then move it to running queue This adds the pipelineRuns in the same order as in the list.
func (*Manager) AddToPendingQueue ¶
func (qm *Manager) AddToPendingQueue(repo *v1alpha1.Repository, list []string) error
func (*Manager) InitQueues ¶
func (qm *Manager) InitQueues(ctx context.Context, tekton versioned2.Interface, pac versioned.Interface) error
InitQueues rebuild all the queues for all repository if concurrency is defined before reconciler started reconciling them.
func (*Manager) QueuedPipelineRuns ¶
func (qm *Manager) QueuedPipelineRuns(repo *v1alpha1.Repository) []string
func (*Manager) RemoveAndTakeItemFromQueue ¶
func (qm *Manager) RemoveAndTakeItemFromQueue(repo *v1alpha1.Repository, run *tektonv1.PipelineRun) string
func (*Manager) RemoveFromQueue ¶
func (*Manager) RemoveRepository ¶
func (qm *Manager) RemoveRepository(repo *v1alpha1.Repository)
func (*Manager) RunningPipelineRuns ¶
func (qm *Manager) RunningPipelineRuns(repo *v1alpha1.Repository) []string
func (*Manager) Snapshot ¶ added in v0.50.0
Snapshot returns the state of every queue, keyed by "namespace/repository". The bool is false when the manager is busy, in which case there is no snapshot to report.
Every concurrency bug so far has been the in-memory queue drifting away from what is really in the cluster. Reading the queue directly turns "nothing is running and I do not know why" into a named PipelineRun.
This takes the same lock the reconciler needs to admit and release runs, so it only ever tries. A diagnostic that can stall the thing it is diagnosing is worse than one that occasionally says "ask again".
type ManagerInterface ¶
type ManagerInterface interface {
InitQueues(ctx context.Context, tekton tektonVersionedClient.Interface, pac versioned.Interface) error
RemoveRepository(repo *v1alpha1.Repository)
QueuedPipelineRuns(repo *v1alpha1.Repository) []string
RunningPipelineRuns(repo *v1alpha1.Repository) []string
AddListToRunningQueue(repo *v1alpha1.Repository, list []string) ([]string, error)
AddToPendingQueue(repo *v1alpha1.Repository, list []string) error
RemoveFromQueue(repoKey, prKey string) bool
RemoveAndTakeItemFromQueue(repo *v1alpha1.Repository, run *tektonv1.PipelineRun) string
}
Source Files