File sent. I was not able to delete the recreate the devices, but I believe I did capture at least one dropout event. If you need a capture with only one device, let me know and I will try to do that overnight. Thanks.
Quick question. Just to satisfy my curiosity, I’m running some tests today to eliminate the VPN from the equation. As part of the testing, I took one of my controllers, changed the IP address, and plugged it into the same switch with the server. After changing the address in the driver, it came up and started running like a champ, as you would expect. I am, however, getting occasional “Unknown” quality overlays on the data from this site. I did a search of the online manual, but it didn’t really explain what triggers, say, “Unknown” versus say “OPC Poor Quality” for example. Can someone enlighten me on this, so when my customer asks, I can have an actual answer for them? 
Latest update:
To test various configurations, I made the following changes:
One controller was temporarily relocated and connected directly to the server’s switch, re-addressed, and is being polled correctly. The only issue is the above mentioned “Unknown” overlay popping up occasionally.
One site was reconfigured so the router did a port-forward for EIP, device settings re-addressed, and site is being polled with the same number of dropouts.
One controller was re-addressed and connected directly to the cable modem (no router in between). This site is seeing the same number of dropouts.
To me, this appears to be some sort of issue on the cable internet side of things. I have verified that the dropouts are shorter for the directly connected controller and the port-forwarded controller. I assume this has to due with the time required to re-establish the VPN tunnel after a dropout?
I’m trying to schedule a meeting with the folks at the cable company to see what we can check on their end, but unless someone else has any other observations or ideas, I don’t know what else it could be.
Update:
It’s looking more and more as if this is a problem on the cable provider’s end. Just to satisfy my curiosity, I started adjusting my OPC-UA Minimum Sampling Interval from 10000 mSec down to 500 mSec. The faster I set the MSI, the fewer dropouts I noticed. I also adjusted my device timeout settings lower as I went. At a MSI of 500 mSec, a State Threshold of 6, and device Read/Write/Comm timeouts of 1500 mSec, my dropouts have probably dropped to a sixth of what they were. Is anyone familiar enough with cable modem systems to guess if they may have some setting in their head end that drops “inactive” TCP connecting after a few seconds? I’m still trying to get a meeting scheduled with their people, and I’d like to go in as well prepared as I can. Thanks.
I’m not sure if this applies, but is something to try.
support.microsoft.com/default.as … -us;283165
Also looking at the logs, Ignition is seeing timeouts.