Forum Discussion
What_Lies_Bene1
Oct 01, 2013Cirrostratus
Yes, that's to be expected. I guess as an alternative you could greatly reduce the TCP connection timeout client-side in the hope that connections are closed (and therefore re-established) more often or worse yet, drop everything down to HTTP/1.0 and/or disable HTTP pipelining. Either way, you are massively increasing your protocol overhead and this isn't desirable.
It would be much better to instruct users to close and restart their browsers or alternatively, only make changes out of hours. Are these options for you?