distribution/storage/notifications/event.go

143 lines
4.9 KiB
Go
Raw Normal View History

package notifications
import (
"fmt"
"time"
"github.com/docker/distribution/digest"
)
// EventAction constants used in action field of Event.
const (
EventActionPull = "pull"
EventActionPush = "push"
EventActionDelete = "delete"
)
// EventTargetType constants used in Target section of Event.
const (
EventTargetTypeManifest = "manifest"
EventTargetTypeBlob = "blob"
)
// EventsMediaType is the mediatype for the json event envelope. If the Event,
// ActorRecord, SourceRecord or Envelope structs change, the version number
// should be incremented.
const EventsMediaType = "application/vnd.docker.distribution.events.v1+json"
// Envelope defines the fields of a json event envelope message that can hold
// one or more events.
type Envelope struct {
// Events make up the contents of the envelope. Events present in a single
// envelope are not necessarily related.
Events []Event `json:"events,omitempty"`
}
// TODO(stevvooe): The event type should be separate from the json format. It
// should be defined as an interface. Leaving as is for now since we don't
// need that at this time. If we make this change, the struct below would be
// called "EventRecord".
// Event provides the fields required to describe a registry event.
type Event struct {
// ID provides a unique identifier for the event.
Implement notification endpoint webhook dispatch This changeset implements webhook notification endpoints for dispatching registry events. Repository instances can be decorated by a listener that converts calls into context-aware events, using a bridge. Events generated in the bridge are written to a sink. Implementations of sink include a broadcast and endpoint sink which can be used to configure event dispatch. Endpoints represent a webhook notification target, with queueing and retries built in. They can be added to a Broadcaster, which is a simple sink that writes a block of events to several sinks, to provide a complete dispatch mechanism. The main caveat to the current approach is that all unsent notifications are inmemory. Best effort is made to ensure that notifications are not dropped, to the point where queues may back up on faulty endpoints. If the endpoint is fixed, the events will be retried and all messages will go through. Internally, this functionality is all made up of Sink objects. The queuing functionality is implemented with an eventQueue sink and retries are implemented with retryingSink. Replacing the inmemory queuing with something persistent should be as simple as replacing broadcaster with a remote queue and that sets up the sinks to be local workers listening to that remote queue. Metrics are kept for each endpoint and exported via expvar. This may not be a permanent appraoch but should provide enough information for troubleshooting notification problems. Signed-off-by: Stephen J Day <stephen.day@docker.com>
2015-01-28 08:27:46 +01:00
ID string `json:"id,omitempty"`
// Timestamp is the time at which the event occurred.
Timestamp time.Time `json:"timestamp,omitempty"`
// Action indicates what action encompasses the provided event.
Action string `json:"action,omitempty"`
// Target uniquely describes the target of the event.
Target struct {
// Type should be "manifest" or "blob"
Type string `json:"type,omitempty"`
// Name identifies the named repository.
Name string `json:"name,omitempty"`
// Digest should identify the object in the repository.
Digest digest.Digest `json:"digest,omitempty"`
// Tag is present if the operation involved a tagged manifest.
Tag string `json:"tag,omitempty"`
// URL provides a link to the content on the relevant repository instance.
URL string `json:"url,omitempty"`
} `json:"target,omitempty"`
// Actor specifies the agent that initiated the event. For most
// situations, this could be from the authorizaton context of the request.
Actor ActorRecord `json:"actor,omitempty"`
// Source identifies the registry node that generated the event. Put
// differently, while the actor "initiates" the event, the source
// "generates" it.
Source SourceRecord `json:"source,omitempty"`
}
// ActorRecord specifies the agent that initiated the event. For most
// situations, this could be from the authorizaton context of the request.
Implement notification endpoint webhook dispatch This changeset implements webhook notification endpoints for dispatching registry events. Repository instances can be decorated by a listener that converts calls into context-aware events, using a bridge. Events generated in the bridge are written to a sink. Implementations of sink include a broadcast and endpoint sink which can be used to configure event dispatch. Endpoints represent a webhook notification target, with queueing and retries built in. They can be added to a Broadcaster, which is a simple sink that writes a block of events to several sinks, to provide a complete dispatch mechanism. The main caveat to the current approach is that all unsent notifications are inmemory. Best effort is made to ensure that notifications are not dropped, to the point where queues may back up on faulty endpoints. If the endpoint is fixed, the events will be retried and all messages will go through. Internally, this functionality is all made up of Sink objects. The queuing functionality is implemented with an eventQueue sink and retries are implemented with retryingSink. Replacing the inmemory queuing with something persistent should be as simple as replacing broadcaster with a remote queue and that sets up the sinks to be local workers listening to that remote queue. Metrics are kept for each endpoint and exported via expvar. This may not be a permanent appraoch but should provide enough information for troubleshooting notification problems. Signed-off-by: Stephen J Day <stephen.day@docker.com>
2015-01-28 08:27:46 +01:00
// Data in this record can refer to both the initiating client and the
// generating request.
type ActorRecord struct {
// Name corresponds to the subject or username associated with the
// request context that generated the event.
Name string `json:"name,omitempty"`
// Addr contains the ip or hostname and possibly port of the client
// connection that initiated the event.
Addr string `json:"addr,omitempty"`
Implement notification endpoint webhook dispatch This changeset implements webhook notification endpoints for dispatching registry events. Repository instances can be decorated by a listener that converts calls into context-aware events, using a bridge. Events generated in the bridge are written to a sink. Implementations of sink include a broadcast and endpoint sink which can be used to configure event dispatch. Endpoints represent a webhook notification target, with queueing and retries built in. They can be added to a Broadcaster, which is a simple sink that writes a block of events to several sinks, to provide a complete dispatch mechanism. The main caveat to the current approach is that all unsent notifications are inmemory. Best effort is made to ensure that notifications are not dropped, to the point where queues may back up on faulty endpoints. If the endpoint is fixed, the events will be retried and all messages will go through. Internally, this functionality is all made up of Sink objects. The queuing functionality is implemented with an eventQueue sink and retries are implemented with retryingSink. Replacing the inmemory queuing with something persistent should be as simple as replacing broadcaster with a remote queue and that sets up the sinks to be local workers listening to that remote queue. Metrics are kept for each endpoint and exported via expvar. This may not be a permanent appraoch but should provide enough information for troubleshooting notification problems. Signed-off-by: Stephen J Day <stephen.day@docker.com>
2015-01-28 08:27:46 +01:00
// Host is the externally accessible host name of the registry instance,
// as specified by the http host header on incoming requests.
Host string `json:"host,omitempty"`
// RequestID uniquely identifies the registry request that generated the
// event.
RequestID string `json:"requestID,omitempty"`
// TODO(stevvooe): Look into setting a session cookie to get this
// without docker daemon.
// SessionID
// TODO(stevvooe): Push the "Docker-Command" header to replace cookie and
// get the actual command.
// Command
}
// SourceRecord identifies the registry node that generated the event. Put
// differently, while the actor "initiates" the event, the source "generates"
// it.
type SourceRecord struct {
// Addr contains the ip or hostname and the port of the registry node
// that generated the event. Generally, this will be resolved by
// os.Hostname() along with the running port.
Addr string `json:"addr,omitempty"`
Implement notification endpoint webhook dispatch This changeset implements webhook notification endpoints for dispatching registry events. Repository instances can be decorated by a listener that converts calls into context-aware events, using a bridge. Events generated in the bridge are written to a sink. Implementations of sink include a broadcast and endpoint sink which can be used to configure event dispatch. Endpoints represent a webhook notification target, with queueing and retries built in. They can be added to a Broadcaster, which is a simple sink that writes a block of events to several sinks, to provide a complete dispatch mechanism. The main caveat to the current approach is that all unsent notifications are inmemory. Best effort is made to ensure that notifications are not dropped, to the point where queues may back up on faulty endpoints. If the endpoint is fixed, the events will be retried and all messages will go through. Internally, this functionality is all made up of Sink objects. The queuing functionality is implemented with an eventQueue sink and retries are implemented with retryingSink. Replacing the inmemory queuing with something persistent should be as simple as replacing broadcaster with a remote queue and that sets up the sinks to be local workers listening to that remote queue. Metrics are kept for each endpoint and exported via expvar. This may not be a permanent appraoch but should provide enough information for troubleshooting notification problems. Signed-off-by: Stephen J Day <stephen.day@docker.com>
2015-01-28 08:27:46 +01:00
// InstanceID identifies a running instance of an application. Changes
// after each restart.
InstanceID string `json:"instanceID,omitempty"`
}
var (
// ErrSinkClosed is returned if a write is issued to a sink that has been
// closed. If encountered, the error should be considered terminal and
// retries will not be successful.
ErrSinkClosed = fmt.Errorf("sink: closed")
)
// Sink accepts and sends events.
type Sink interface {
// Write writes one or more events to the sink. If no error is returned,
// the caller will assume that all events have been committed and will not
// try to send them again. If an error is received, the caller may retry
// sending the event. The caller should cede the slice of memory to the
// sink and not modify it after calling this method.
Write(events ...Event) error
// Close the sink, possibly waiting for pending events to flush.
Close() error
}