Publish block when over the memory / disk threshold

Options
Elaine
Elaine Member Posts: 1
Hi, we are migrating from rabbit to solace now.
There is a feature (https://www.rabbitmq.com/alarms.html) RabbitMQ will block connections that publish messages when over the memory/disk threshold.

I see there is a way to set the threshold in solace: https://docs.solace.com/Monitoring/Configuring-System-Event-Thresholds.htm

But are there similar feature that can block publication or notify client in solace? any guideline?

Thanks in advance

Best Answer

  • uherbst
    uherbst Member, Employee Posts: 121 Solace Employee
    #2 Answer ✓
    Options

    Hi @Elaine ,

    About memory thresholds: This is not needed, because the Solace broker pod requires all the memory during startup that is needed at maximum for operations.

    About disk threshold:
    I'm assuming, you're talking about persistent messaging (Solace calls this "guaranteed messaging") where each message is persisted to disk before broker sends back an ACK to the publisher.

    If disk space is exhausted and the message can't be persisted to disk, there will be a negative ACK back to the publisher to inform the publisher about that condition.

    There is another "special case" for it:

    If overall disk space is good, but one (of possible multiple) queues, that would receive that message, are over queue-quota, you can set "reject-msg-to-sender-on-discard" to inform the publisher. (for details see here: https://solace.com/blog/pubsub-message-handling-features-reject-message-to-sender-on-discard/ )

    Does that answer your questions ?

    Uli

Answers

  • uherbst
    uherbst Member, Employee Posts: 121 Solace Employee
    #3 Answer ✓
    Options

    Hi @Elaine ,

    About memory thresholds: This is not needed, because the Solace broker pod requires all the memory during startup that is needed at maximum for operations.

    About disk threshold:
    I'm assuming, you're talking about persistent messaging (Solace calls this "guaranteed messaging") where each message is persisted to disk before broker sends back an ACK to the publisher.

    If disk space is exhausted and the message can't be persisted to disk, there will be a negative ACK back to the publisher to inform the publisher about that condition.

    There is another "special case" for it:

    If overall disk space is good, but one (of possible multiple) queues, that would receive that message, are over queue-quota, you can set "reject-msg-to-sender-on-discard" to inform the publisher. (for details see here: https://solace.com/blog/pubsub-message-handling-features-reject-message-to-sender-on-discard/ )

    Does that answer your questions ?

    Uli