What happens to replay logs when using Dynamic Message Routing?

toliver

If I'm using Dynamic Message Routing/ an event mesh and have many Solace instances connected together, what happens with the replay logs? Does every instance have a complete replay log or is it distributed across all instances?

Is there any way to add redundancy to replay logs? https://docs.solace.com/Overviews/Message-Replay-Overview.htm This page says that the replay log cannot be guaranteed to be replicated if using replication

himanshu

Hi @toliver - first of all, great question. Replay and DMR are both powerful features and one should really understand how they work, especially together.

Any message that makes it to a message VPN in PubSub+ broker as a Guaranteed Message, it will make it to the Replay Log.

DMR supports both Direct Messaging and Guaranteed Messaging. When you use DMR, if a message makes it to a remote VPN where there is a queue subscribing to a topic, it is promoted to Guaranteed Messaging and hence, it will make it to Replay log. If the message is attracted to a remote VPN due to a direct topic subscription (no queues involved), then it will not make it to Replay log.

Similarly, we have replication or DR. With DR, publishers publish messages to primary broker which are then replicated to the DR site. Once the messages makes it to the DR site, they will then be recorded in the Replay log. However, Solace's DR is very efficient. If the message is consumed by a consumer from the primary broker 'quickly enough', it will not replicate that message to the DR site. This is a great feature because, in case of a DR event, you want to failover to DR site and pick up where you left off. However, this also means that there will be some messages which will not make it to the Replay log on the DR site.

If this is a feature that you are interested in, let us know what your usecase is and which company you work at so we can add it to our feature enhancement list.

Hope this helps,
Himanshu

RunningRoach

@Aaron : Are we saying that Replay is not HA compatible, or that the default HA settings for replication do not take into account Replay logs and some settings have to be changed?