We help IT Professionals succeed at work.

Cisco ACE Load Balancer

sumit2012
sumit2012 asked
on
Lets suppose if i have configured a probe of 10 sec and inteval is 5. That means 10 sec * 5 = 50 sec , is the time if by that time the probe didn't get the response it will remove the server from the serverfarm.

My question is what happens b/w the first probe failure to 50 sec time....During that period of time is the client request i mean the server still receives the client request before moving out from the serverfarm..
Comment
Watch Question

Commented:
If the probe is every 10 seconds then this is the interval. 5 is ? - let's say this is the fail detect count. So it could be up to 5 *10 + 9 = 59 seconds that this server is marked out of service.
During this time, if the service has failed, any client connections to this server will also fail and timeout.

Author

Commented:
when we configure the probe. we need to define like fail-detect is lets suppose 10 second and the interval lets suppose i have defined 5.

So the fail-detect total time is = 50 seconds thats means if the probe will not get any response with in 50 seconds then it will put that rserver out from the serverfarm.

My question is that when the probe gets failed to very frist time to till 50 seconds. Is that rserver still having any questions or all connections gets drops as soon as the very first probe gets failed but before completing the total time defined in the fail-detect. So its only remove that rserver from serving the client request on the completion 50 seconds of fail-detect.

Please correct if my understanding is wrong.

Commented:
Just to get the terminology right, the interval (seconds) is how often a probe is performed, the fail detect is a count and is the consecutive number of failed probe resonses required to indicate a loss of service.

The answer to your question is yes.

Author

Commented:
You are correct. but my question was still not answered.

This is correct that interval means how often the probe is performed. Lets suppose the probe is to be performed 5 times. So till the completion of 5 times the probing gets complete then only the rserver is make as OUT OF SERVICE. or from the very first time when the probe gets failed. The client request stops going to that particular rserver from the very first failed probe.

Commented:
Thank you.

"Lets suppose the probe is to be performed 5 times" - OK

"The client request stops going to that particular rserver from the very first failed probe." - No
"So till the completion of 5 times the probing gets complete then only the rserver is make as OUT OF SERVICE" - Yes

Author

Commented:
As you said above :-

"The client request stops going to that particular rserver from the very first failed probe." - No

Now the question is will there be no any connections drops to the application.
Commented:
The answer is in my first comment. If the service on the rserver has failed, then any existing or new client connections will timeout, that is, drop.

If there is a concern with the delay of failed rservers being taken out of service, there is nothing stopping you from reducing the interval and fail detect count to values that suit you. But you don't want the settings to be too sensitive so that a temporary delay in response causes the rservers to be taken out of service unnecessarily.

Explore More ContentExplore courses, solutions, and other research materials related to this topic.