Since moving to BP V7 we have has significant problems with processes becoming stuck at the Attach stage, and then going to Warning in the Control Room. In some cases they don't error or fail, the process does not terminate so we have to log in and force the process offline before we can restart it. Over the last couple of weeks we have found several common triggers and you should consider each of these.
Process - Object relationships: A single process can call multiple objects, that continued to work. We saw a number of issues where the objects were then calling child objects. The common error was 'more than one application matched'. We either moved the steps from the child object to the parent, or re-mapped the child object to attach the the browser using win32 and avoiding title names.
Wait before attach: when calling an object a common practice for us is to start each page with an attach. In the attach page after the IsConnected? check, if the answer is False, feed into a 5s wait before calling your attach.
Wait after attach: occasionally, after getting attached, the first action the object took would fail. By adding a 5s wait after the attach it gave time for the extension to lock in properly.
BP Timeout Settings (Advanced): after the steps above we noticed that some automations continued to fail. Deep diving these we saw that the MessagingHost process was taking a long time to start. We have adjusted our timers in BP > System > System > Timeouts, configuring the top four options to 600s (10 mins). This extends the time that BP will allow for the browser to start, extension to fire up, MessagingHost to start, and then get connected. Although we only made this change today, it appears to have had immediate benefit. Note that this is a maximum setting, so BP will only wait as long as it needs, not for the full 10 mins.
