4.1. Discovery

4.1.1. SPDP2 participants may not have completed discovery if IP mobility event occurred during discovery

When using Simple Participant Discovery Protocol 2.0, discovery may not have completed between two DomainParticipants if one DomainParticipant changed locators due to an IP mobility event before its configuration message was received by the remote participant. You would have had to wait for DomainParticipant liveliness to expire at the participant_liveliness_lease_duration for discovery to be restarted. Now, the locator change is correctly propagated to the remote participant and participant discovery will complete.

[RTI Issue ID CORE-13384]

4.1.2. Crash if initial_peers sequence contained a NULL string

Previously, if you configured the initial peers sequence through code, you could potentially add a NULL element. Connext did not check for the NULL element; therefore, when the DomainParticipant was created in this case, Connext crashed. Now a NULL element will be reported, resulting in an ‘inconsistent qos’ failure when creating the DomainParticipant.

[RTI Issue ID CORE-13802]

4.1.3. Failure to deserialize participant discovery information incorrectly allowed discovery to complete

It was possible for participant discovery to “succeed” even if the deserialization of the participant discovery information failed. In those cases, this error was printed:

PRESCstReaderCollator_storeSampleData:deserialize sample error in topic 'DISCParticipant' with type 'DISCParticipantParameter'

This incorrect ‘success’ could have led to unexpected behavior or crashes. This problem has been fixed. Now participant discovery won’t complete if deserialization issues are detected.

[RTI Issue ID CORE-12952]

4.1.4. Unbounded memory growth when creating/deleting DomainParticipants

In Connext 7.1.0, a DomainParticipant was not freeing some of the memory associated with a remote DomainParticipant that was deleted. This may have led to unbounded memory growth if your applications continuosly create/delete DomainParticipants. When this growth occurred, you may have seen the following error message:

ERROR [DELETE DP|LC:DISC]COMMENDAnonWriterService_assertRemoteReader:DELETION FAILURE | skiplist node already removed

This problem has been fixed.

[RTI Issue ID CORE-13964]