Not sure if this is the proper place to ask this or not, but I have an odd issue with a VM (really multiple VMs but they all do the same thing). The VMs run RHEL 6.5 and spool a print job to a printer at the far end of a WAN connection. The printer doesn't handle out of order packets correctly and ultimately just waits for the OS to decide it should resend things and restart. This causes major delays in the print job. The app was migrated from a RHEL5.10 physical system to the VM where the problem started. The printer vendor has admitted a problem in the way they handle out of order packets but isn't fixing it in the printer(s) we have. When the app was on RHEL5/physical system the problem did not occur. At least, it wasn't severe enough for us to look really deep into. The application team wants to move back to RHEL5/physical until the problem is resolved but I am reluctant to do that because it effectively means they won't migrate into a VM for another 1-2 years.
I think a previously existing problem is being made worse by the VMware layer, but don't know what I can do about it. We run the Nexus 1000V with LACP back to redundant Nexus 7000's before the connection routes to the WAN circuit. I have been told the problem can be reproduced without going thru the WAN, and we have tried using the VSS on a separate pair of cables, but it didn't help. Can anyone think of anything that I can check or tweak to make things easier for my end users? I really don't want to move back to physical. It makes so many processes 10x more difficult.
We are running ESXi 5.5u2 with vSphere 5.5u2 on IBM 3850m3 servers. They aren't memory, CPU, or disk constrained at all as far as anyone can tell, and the network from the box isn't saturated or that busy at all.
Thanks.
Tom
