our labs server has failed and people are working to get it going…meanwhile our pending transactions to the lab are rapidly filling up. is there a limit to how much it will hold and how do you send to a file and resend from that. we are using cloverleaf 5.4.1 on a unix box.
Your essentially limited by disk space, system resources, and the size of the recovery database (2GB). I would just let the engine continue to queue the messages and monitor the situation. Personally, I think it’s better to let the engine manage the transactions, rather than write them off to a file and then manually send them. We’ve had systems down where we’ve had 30,000+ messages queued and just let the engine do it’s work. If you write them to a file, then resend, you’re not really saving anything – you are still consuming disk by writing to the file and then you’ll consume engine resources with a big ‘resend’ when it comes time to do so.
I would just let the engine do it’s job while keeping an eye on things.
I’d agree with Jim. If you don’t have some processes and procedures thought out an in place prior to the event, you should just leave well enough alone.
Stopping a thread with thousands of messages queued up will just cause the entire process to lock up when you start the thread up again under its new configuration (say a disk file) as the thread loads those state 11 messages back in, and then processes all of the state 7 messages queued in the translate thread on to the destination thread. This could cause inbound threads in the process to essentially timeout on their responses to their connections. It’s a huge mess unless, as I said, you are ready for it.
.
.