When comparing with active users on Google Analytics, Galaxy connections is always ~3x higher. I have mixmax:smart-disconnect installed, so either it’s not working or what counts as a connection is dramatically different than what counts as an “active user”.
Idle users drop off of GA count… whereas they remain in the Galaxy connection list as long as the browser is open.
For performance, we do a disconnect on the client if the client is idle for 8 minutes. If the user moves the mouse or types we reconnect. As soon as we added this code our peak load cut in half.
Thanks for the reply. I’ve been using this package (which is supposed to do what you just described) but I’m not sure it’s even working at this point.
I’m not familiar with that package, but it sound awesome. 2 ways to find out if it works.
- Log connections when running on localhost and see if it works.
- Otherwise, watch the galaxy dashboard on your sandbox/staging server.
In both case… since you have controlled environment you can see what happens to the connection count as you manipulate the client.