Greetings,
I have a case of couple of nodes behind a cluster. There is randomly
allocated backlog data. Most of the times load balancer hits correct node
with required piece of data. But when it hits node without it - then it's
bad. I don't want to use NFS as single point of failure.
Initial thought was to send a request message to an ActiveMQ topic, every
node checked if they had required piece of data and only the one possessing
it - only then send reply to internally created temporary queue. The
problem is that i didn't manage to prevent sending messages to temporary
queue when required piece of data was not found.
I've tried with creating random name queue from code directly, but it
doesn't get removed from broker.
Any thoughts?
Best regards,
*Serge*
I have a case of couple of nodes behind a cluster. There is randomly
allocated backlog data. Most of the times load balancer hits correct node
with required piece of data. But when it hits node without it - then it's
bad. I don't want to use NFS as single point of failure.
Initial thought was to send a request message to an ActiveMQ topic, every
node checked if they had required piece of data and only the one possessing
it - only then send reply to internally created temporary queue. The
problem is that i didn't manage to prevent sending messages to temporary
queue when required piece of data was not found.
I've tried with creating random name queue from code directly, but it
doesn't get removed from broker.
Any thoughts?
Best regards,
*Serge*