Edit

Azure Service Bus - Message expiration (Time to Live)

  • Article

The payload in a message, or a command or inquiry that a message conveys to a receiver, is almost always subject to some form of application-level expiration deadline. After such a deadline, the content is no longer delivered, or the requested operation is no longer executed.

For development and test environments in which queues and topics are often used in the context of partial runs of applications or application parts, it's also desirable for stranded test messages to be automatically garbage collected so that the next test run can start clean.

Note

If you aren't familiar with Service Bus concepts yet, see Service Bus concepts, and Service Bus queues, topics, and subscriptions.

The expiration for any individual message can be controlled by setting the time-to-live system property, which specifies a relative duration. The expiration becomes an absolute instant when the message is enqueued into the entity. At that time, the expires-at-utc property takes on the value enqueued-time-utc + time-to-live. The time-to-live (TTL) setting on a brokered message isn't enforced when there are no clients actively listening.

Past the expires-at-utc instant, messages become ineligible for retrieval. The expiration doesn't affect messages that are currently locked for delivery. Those messages are still handled normally. If the lock expires or the message is abandoned, the expiration takes immediate effect. While the message is under lock, the application might be in possession of a message that has expired. Whether the application is willing to go ahead with processing or chooses to abandon the message is up to the implementer.

Extremely low TTL in the order of milliseconds or seconds might cause messages to expire before receiver applications receive it. Consider the highest TTL that works for your application.

Note

For scheduled messages, you specify the enqueue time at which you want the message to materialize in the queue for retrieval. The time at which the message is sent to Service Bus is different from the time at which the message is enqueued. The message expiration time depends on the enqueued time, not on the time at which the message is sent to Service Bus. Therefore, the expires-at-utc is still enqueued time + time-to-live.

For example, if you set the ScheduledEnqueueTimeUtc to 5 minutes from UtcNow, and TimeToLive to 10 minutes, the message will expire after 5 + 10 = 15 minutes from now. The message materializes in the queue after 5 minutes and the 10 minute counter starts from then.

Entity-level expiration

All messages sent into a queue or topic are subject to a default expiration that is set at the entity level. It can also be set in the portal during creation and adjusted later. The default expiration is used for all messages sent to the entity where time-to-live isn't explicitly set. The default expiration also functions as a ceiling for the time-to-live value. Messages that have a longer time-to-live expiration than the default value are silently adjusted to the default message time-to-live value before being enqueued.

Note

  • The default time-to-live value for a brokered message is the largest possible value for a signed 64-bit integer if not otherwise specified.
  • For messaging entities (queues and topics), the default expiration time is also largest possible value for a signed 64-bit integer for Service Bus standard and premium tiers. For the basic tier, the default (also maximum) expiration time is 14 days.
  • If the topic specifies a smaller TTL than the subscription, the topic TTL is applied.

Expired messages can optionally be moved to a dead-letter queue. You can configure this setting programmatically or using the Azure portal. If the option is left disabled, expired messages are dropped. Expired messages moved to the dead-letter queue can be distinguished from other dead-lettered messages by evaluating the dead-letter reason property that the broker stores in the user properties section.

If the message is protected from expiration while under lock and if the flag is set on the entity, the message is moved to the dead-letter queue as the lock is abandoned or expires. However, it isn't moved if the message is successfully settled, which then assumes that the application has successfully handled it, in spite of the nominal expiration. For more information about message locks and settlement, see Message transfers, locks, and settlement.

The combination of time-to-live and automatic (and transactional) dead-lettering on expiry are a valuable tool for establishing confidence in whether a job given to a handler or a group of handlers under a deadline is retrieved for processing as the deadline is reached.

For example, consider a web site that needs to reliably execute jobs on a scale-constrained backend, and which occasionally experiences traffic spikes or wants to be insulated against availability episodes of that backend. In the regular case, the server-side handler for the submitted user data pushes the information into a queue and subsequently receives a reply confirming successful handling of the transaction into a reply queue. If there's a traffic spike and the backend handler can't process its backlog items in time, the expired jobs are returned on the dead-letter queue. The interactive user can be notified that the requested operation takes a little longer than usual, and the request can then be put on a different queue for a processing path where the eventual processing result is sent to the user by email.

Expiration for session-enabled entities

For session-enabled queues or topics' subscriptions, messages are locked at the session level. If the TTL for any of the messages expires, all messages related to that session are either dropped or dead-lettered based on the dead-lettering enabled on messaging expiration setting on the entity. In other words, if there's a single message in the session that has passed the TTL, all the messages in the session are expired. The messages expire only if there's an active listener.

Temporary entities

Service Bus queues, topics, and subscriptions can be created as temporary entities, which are automatically removed when they haven't been used for a specified period of time.

Automatic cleanup is useful in development and test scenarios in which entities are created dynamically and aren't cleaned up after use, due to some interruption of the test or debugging run. It's also useful when an application creates dynamic entities, such as a reply queue, for receiving responses back into a web server process, or into another relatively short-lived object where it's difficult to reliably clean up those entities when the object instance disappears.

The feature is enabled using the auto delete on idle property on the namespace. This property is set to the duration for which an entity must be idle (unused) before it's automatically deleted. The minimum value for this property is 5 minutes.

Important

Setting the Azure Resource Manager lock-level to CanNotDelete, on the namespace or at a higher level doesn't prevent entities with AutoDeleteOnIdle from being deleted. If you don't want the entity to be deleted, set the AutoDeleteOnIdle property to DataTime.MaxValue.

Idleness

Here's what considered idleness of entities (queues, topics, and subscriptions):

Entity What's considered idle
Queue
  • No sends
  • No receives
  • No updates to the queue
  • No scheduled messages
  • No browse/peek
Topic
  • No sends
  • No updates to the topic
  • No scheduled messages
  • No operations on the topic's subscriptions (see the next row)
Subscription
  • No receives
  • No updates to the subscription
  • No new rules added to the subscription
  • No browse/peek

Important

If you have auto forwarding setup on the queue or subscription, that is equivalent to having a receiver peform receives on the queue or subscription and they will not be idle.

SDKs

Next steps

To learn more about Service Bus messaging, see the following articles: