Thread/Process sending large msgs hangs if ACK not received

[Tim Malaney]

Since recently upgrading to 6.0.2 in a Linux environment have experienced the following issue. Sending large result messages (2-4MB) that contain an embedded pdf. if the receiving systems is slow with an ack (at least we think this is the cause) and more the a a couple messages get queued,  we get the error below and the thread does not attempt to resend the message.

[pdl :PDL :ERR /0:HPF_MUSE_RECV:11/18/2014 09:54:06] write timeout expired

[pdl :PDL :ERR /0:HPF_MUSE_RECV:11/18/2014 09:54:06] PDL signaled exception: code 1, msg write failure

Thread shows as UP with no additional activity except messages continue to queue. If we stop the thread the process goes down and produces a core dump. Starting the process just puts it back to the same state of doing nothing. We end up taking the process down, clearing the recovery DB and manually sending the messages one at a time. Timeout is set to 120 seconds and when running normal we don’t see more then a couple second lag getting an ack back.  No issues with this application with 5.8. Anyone see anythings similar?

Thank

Topic

[Kevin Crist]

We moved to 6.0.1 on Linux a month and a half ago and started seeing these almost immediately with our embedded pdfs going outbound. It usually takes restarting the process a couple of times and restarting the interface it’s going to to get it running again. Would happen about once, maybe twice a week. I am experimenting with message throttling just for S&Gs. It is interesting that just today i got the same error on an ADT feed, which led me here.

[Tim Malaney]

We may go to throttling but as a last resort. We are worried we would be masking a problem that could show up in another form. We never see

RESEND_OB_MSG -No response within timeout or state 14 message resent.

RESEND_OB_MSG – Resending msg at……

[Author]

Replies