Hi, we're developing a monitoring system that needs to consider all new values in tags. We're using event pipe to obtain these new values from PI and AF. The clients of the system can, at any time, include new tags or exclude existing ones to be monitored. So, the list of tags subscribed to be listened in event pipe can change. About this scenario, we have two problems: 1) After some time running, the system locks. Using profile tools, we identify that there is a deadlock in SDK, caused by a tentative of a connection x a broadcast disconnection event, sent by an internal class of SDK. Do you have some information about this problem? Is there some known problem about this strategy of changing the list of subscribed tags while event pipe are running? 2) As a monitoring system we need to have some failover strategy. We can't lose tag values. So we need a high availability environment to event pipe. Do you have some strategy, pattern or architecture to get this? We tried to mantain the timestamp of the last get from event pipe, and load all tag values from it until now. But this strategy caused an long time to execute that we can't pay.