For true guaranteed delivery with durable subscribers, the client must use CLIENT_ACKNOWLEDGE or a transacted session. AUTO_ACKNOWLEDGE means the server removes the message as soon as it’s sent to the client, before the client processes it. If the client disconnects at that moment, the message is lost because it’s no longer on the server and wasn’t processed. The other options relate to different aspects of reliability but are not the root cause of loss with a disconnected durable subscriber.

Store and Forward, when enabled for durable subscriptions, persists messages to disk to ensure delivery to offline subscribers. If there are many durable subscribers or a high message rate, this can lead to excessive disk I/O, causing the observed delays. While message limits (option D) can cause message discarding, they typically don’t cause sustained high disk I/O. Compression might reduce I/O, and load balancing distributes connections, not message persistence workload.

For clients to automatically fail over to the backup server in an FT pair, they must be configured to use the Fault Tolerant protocol by connecting via an ‘ftl://’ URL (e.g., ftl://server1:port,server2:port). If they connect directly to the primary’s ‘tcp://’ address, they have no inherent knowledge of the backup and will simply lose connection when the primary fails. Reconnect attempts (option C) would just try the same failed ‘tcp://’ address.

The ‘redelivery delay’ property specifies a server-side wait period before a negatively acknowledged (or unacknowledged) message becomes visible on the queue again for any consumer. It is not a client-specific delay (option D). This prevents a fast-failing message from immediately consuming 100% of a consumer’s resources in a tight loop. The message is not immediately redelivered (A) and is not sent to the DMQ (C) on the first failure unless ‘maxRedelivery’ is set to 1.

This is a classic publish/subscribe (pub/sub) scenario. A Topic is the correct destination for one-to-many messaging. Durable subscriptions are crucial here because they guarantee that each microservice (subscriber) will receive the message regardless of whether it is online at the moment the message is published. If a microservice is down, the message is stored and delivered when it reconnects. Non-durable subscribers (B) would lose messages if offline. A single queue (A) would deliver the message to only one consumer. Manually bridging to three queues (D) adds unnecessary complexity and potential failure points.

For non-critical, real-time data where performance is the priority, you minimize overhead. NON_PERSISTENT mode avoids disk I/O. AUTO_ACKNOWLEDGE has less client-server chatter than CLIENT_ACKNOWLEDGE. Transactions add significant overhead and are not needed if message loss is acceptable. DUPS_OK_ACKNOWLEDGE (D) is a variant of AUTO_ACKNOWLEDGE but is less common. PERSISTENT mode (B and D) would add the disk I/O latency you’re trying to avoid.

In the `authorization` section of tibemsd.conf, the syntax ` = , ` is used to group destinations (queues or topics) so that a single permission rule (like `allow receive user=alice`) can be applied to the entire group. This simplifies security management. It does not create routing rules (A), define client-side aliases (B), or enable wildcard consumption (D), which is a client-side concept using destination names like `queue.>`.

TIBCO EMS exposes management and monitoring data through a standard JMX (Java Management Extensions) interface. Queue depth, along with a wealth of other server, connection, and destination statistics, is available via dedicated MBeans. This is the intended, programmatic API for monitoring. Option B is fictitious; QueueBrowser lets you *see* messages, not count them via a single method. Parsing logs (C) or command output (D) is fragile and not a true API.

In SSL/TLS, the truststore (on the client) contains the Certificate Authorities (CAs) or specific certificates that the client trusts. The server presents its certificate during the handshake. The client uses its truststore to verify that the server’s certificate is valid and signed by a trusted authority, preventing man-in-the-middle attacks. The client’s own identity (private key) is in a keystore (A). Credentials (C) are not stored in a truststore. Payload encryption (D) is handled by the TLS protocol itself, not a file.

For true guaranteed delivery with durable subscribers, the client must use CLIENT_ACKNOWLEDGE or a transacted session. AUTO_ACKNOWLEDGE means the server removes the message as soon as it’s sent to the client, before the client processes it. If the client disconnects at that moment, the message is lost because it’s no longer on the server and wasn’t processed. The other options relate to different aspects of reliability but are not the root cause of loss with a disconnected durable subscriber.0