|
RTI Connext C API
Version 6.1.2
|
This QoS policy specifies whether or not RTI Connext will store and deliver previously published data samples to new DDS_DataReader entities that join the network later. More...
Data Fields | |
| DDS_DurabilityQosPolicyKind | kind |
| The kind of durability. More... | |
| DDS_Boolean | direct_communication |
| <<extension>> Indicates whether or not a TRANSIENT or PERSISTENT DDS_DataReader should receive samples directly from a TRANSIENT or PERSISTENT DDS_DataWriter More... | |
| DDS_Long | writer_depth |
| <<extension>> Indicates the number of samples per instance that a durable DDS_DataWriter will send to a late-joining DDS_DataReader. More... | |
This QoS policy specifies whether or not RTI Connext will store and deliver previously published data samples to new DDS_DataReader entities that join the network later.
It is possible for a DDS_DataWriter to start publishing data before all (or any) DDS_DataReader entities have joined the network.
Moreover, a DDS_DataReader that joins the network after some data has been written could potentially be interested in accessing the most current values of the data, as well as potentially some history.
This policy makes it possible for a late-joining DDS_DataReader to obtain previously published samples.
By helping to ensure that DataReaders get all data that was sent by DataWriters, regardless of when it was sent, using this QoS policy can increase system tolerance to failure conditions.
Note that although related, this does not strictly control what data RTI Connext will maintain internally. That is, RTI Connext may choose to maintain some data for its own purposes (e.g., flow control) and yet not make it available to late-joining readers if the DURABILITY policy is set to DDS_VOLATILE_DURABILITY_QOS.
For the purpose of implementing the DURABILITY QoS kind TRANSIENT or PERSISTENT, RTI Connext behaves as if for each Topic that has DDS_DurabilityQosPolicy::kind of DDS_TRANSIENT_DURABILITY_QOS or DDS_PERSISTENT_DURABILITY_QOS there is a corresponding "built-in" DDS_DataReader and DDS_DataWriter configured with the same DURABILITY kind. In other words, it is as if somewhere in the system, independent of the original DDS_DataWriter, there is a built-in durable DDS_DataReader subscribing to that Topic and a built-in durable DataWriter re-publishing it as needed for the new subscribers that join the system. This functionality is provided by the RTI Persistence Service.
The Persistence Service can configure itself based on the QoS of your application's DDS_DataWriter and DDS_DataReader entities. For each transient or persistent DDS_Topic, the built-in fictitious Persistence Service DDS_DataReader and DDS_DataWriter have their QoS configured from the QoS of your application's DDS_DataWriter and DDS_DataReader entities that communicate on that DDS_Topic.
For a given DDS_Topic, the usual request/offered semantics apply to the matching between any DDS_DataWriter in the domain that writes the DDS_Topic and the built-in transient/persistent DDS_DataReader for that DDS_Topic; similarly for the built-in transient/persistent DDS_DataWriter for a DDS_Topic and any DDS_DataReader for the DDS_Topic. As a consequence, a DDS_DataWriter that has an incompatible QoS will not send its data to the RTI Persistence Service, and a DDS_DataReader that has an incompatible QoS will not get data from it.
Incompatibilities between local DDS_DataReader and DDS_DataWriter entities and the corresponding fictitious built-in transient/persistent entities cause the DDS_REQUESTED_INCOMPATIBLE_QOS_STATUS and DDS_OFFERED_INCOMPATIBLE_QOS_STATUS to change and the corresponding Listener invocations and/or signaling of DDS_Condition objects as they would with your application's own entities.
The value of DDS_DurabilityServiceQosPolicy::service_cleanup_delay controls when RTI Persistence Service is able to remove all information regarding a data instances.
Information on a data instance is maintained until the following conditions are met:
1. The instance has been explicitly disposed (instance_state = NOT_ALIVE_DISPOSED),
and
2. All samples for the disposed instance have been acknowledged, including the dispose sample itself,
and
3. A time interval longer that DDS_DurabilityServiceQosPolicy::service_cleanup_delay has elapsed since the moment RTI Connext detected that the previous two conditions were met. (Note: Only values of zero or DDS_DURATION_INFINITE are currently supported for the service_cleanup_delay)
The utility of DDS_DurabilityServiceQosPolicy::service_cleanup_delay is apparent in the situation where an application disposes an instance and it crashes before it has a chance to complete additional tasks related to the disposition. Upon restart, the application may ask for initial data to regain its state and the delay introduced by the service_cleanup_delay will allow the restarted application to receive the information on the disposed instance and complete the interrupted tasks.
The value offered is considered compatible with the value requested if and only if the inequality offered kind >= requested kind evaluates to 'TRUE'. For the purposes of this inequality, the values of DURABILITY kind are considered ordered such that DDS_VOLATILE_DURABILITY_QOS < DDS_TRANSIENT_LOCAL_DURABILITY_QOS < DDS_TRANSIENT_DURABILITY_QOS < DDS_PERSISTENT_DURABILITY_QOS.
| DDS_DurabilityQosPolicyKind DDS_DurabilityQosPolicy::kind |
The kind of durability.
[default] DDS_VOLATILE_DURABILITY_QOS
| DDS_Boolean DDS_DurabilityQosPolicy::direct_communication |
<<extension>> Indicates whether or not a TRANSIENT or PERSISTENT DDS_DataReader should receive samples directly from a TRANSIENT or PERSISTENT DDS_DataWriter
When direct_communication is set to DDS_BOOLEAN_TRUE, a TRANSIENT or PERSISTENT DDS_DataReader will receive samples from both the original DDS_DataWriter configured with TRANSIENT or PERSISTENT durability and the DDS_DataWriter created by the persistence service. This peer-to-peer communication pattern provides low latency between end-points.
If the same sample is received from the original DDS_DataWriter and the persistence service, the middleware will discard the duplicate.
When direct_communication is set to DDS_BOOLEAN_FALSE, a TRANSIENT or PERSISTENT DDS_DataReader will only receive samples from the DDS_DataWriter created by the persistence service. This brokered communication pattern provides a way to guarantee eventual consistency.
[default] DDS_BOOLEAN_TRUE
| DDS_Long DDS_DurabilityQosPolicy::writer_depth |
<<extension>> Indicates the number of samples per instance that a durable DDS_DataWriter will send to a late-joining DDS_DataReader.
The default value, DDS_AUTO_WRITER_DEPTH, makes this parameter equal to the following:
The writer_depth must be <= DDS_HistoryQosPolicy::depth.
writer_depth applies only to non-volatile DataWriters (those for which the kind is TRANSIENT_LOCAL, TRANSIENT, or PERSISTENT).
When DDS_BatchQosPolicy::enable is set to true, writer_depth acts as a minimum number of samples per instance that will be sent to late joiners, as opposed to the maximum. As long as a batch contains a single sample that falls within the writer_depth for the instance to which it belongs, then the entire batch will be sent. This means that batches sent to late-joining DataReaders may contain more samples per instance than is specified by the writer_depth QoS setting.
When a DataWriter responds to a TopicQuery, the samples that are evaluated against the TopicQuery filter are only those samples that fall within the writer_depth, not the DDS_HistoryQosPolicy::depth.
Setting writer_depth on the DataReader side will be ignored.
[default] DDS_AUTO_WRITER_DEPTH