Forum Discussion
CirrusDefault retry time inband monitor
MVPI could see this causing an issue if you have multiple monitors and some sort of rate limiting on the server side. In addition, depending on the monitor, you will be opening a new connection every time you run the monitor. This becomes an even larger issue if you have thousands of health monitors running simultaneously but at low pool member/monitor levels it shouldn't really be an issue.
Cirruswhen the monitor runs simultaneously, will it cause disruption to server side traffic? I think when the monitor runs every 1 second it will bring good performance to find out the server condition faster.
isn't it when the retry becomes 300 seconds, and the monitor is up in less than that time, does it mean that there is ineffectiveness in the use of server resources that are actually available?
- Paulius
If you have enough connections running to the server and the monitor on top of that you might experience issues but most likely not related to just performing the health monitor. I would like to note that most health monitors are not run that tight because sometimes you do have delays in server response which could cause the server to be pulled out of rotation when it really shouldn't be. I would increase this to 5 seconds with 3 retries and an overall timeout of 16 seconds.
- Zen_Y
this becomes interesting when we add the possibility that there is a server delay in responding. thanks for the discussion.
- Zen_Y
yeah, interesting idea.
i just realized that possibility, it could be that the monitor is invalid because some servers may take time to process it. which is actually normal, it will look like a problem.
thanks for the suggestion
