Persistence timeout setting question...


Is there a reason or scenario where having a higher persistence timeout setting (say having it set at 4 hours instead of 30 minutes) on a HTTPS virtual service (scheduling method is Round Robin, persistence option is Super HTTPS) might not be a good idea?

I can't decide if it might increase the propensity of uneven load distribution between the real servers.



Hello ALIC
Yes a higher persistence timeout may be a caused of uneven load distribution, Due to any connection that is made to the loadmaster that has an existing persistence entry will be directed back to the server that client previously connected to.
The persistence timeout value you set, should take into account if you have requirement for user connections to be redirected back to the same server and if so for what time period



I have only been with this client for a couple of days and have been handed this issue.

They use a KEMP Load Master (7.1) and have 3 web servers hosting a web based application allowing clients to view content and also process an online booking.

Some clients are reporting that when they hit the online booking they are sometimes disconnected during the booking and have to log back in. When doing so they are then reconnected to a different server, not the same server.

Q1. would KEMP LM just disconnect a users session without the connection being idle?

Q2. If so, would the session be reconnected to a different server in the 3 server farm?

KEMP Settings

3 real servers hosting 1 online service

Basic Properties/Service Type - HTTP/HTTPS

(Activate/Deactivate service 'Ticked') What does this tick mean?

Standard Options;

Subnet originating requests 'Ticked'

Persistence Options - Super HTTP - Timeout 20 mins

Scheduling Method - Round Robin (No idle timeout specified, no NAT'ing)

QoS - Normal Service


This does not happen to all users, only a handful and just need to eliminate KEMP as the routecause.

In advance thank you for your support.