I am currently setting up a server pool with AWS. This is a simple setup with two database servers with a scalable array of servers and two load balancers before all this. Each machine has a transition to another resource, and everything should be quite reliable.
Load balancers should be able to switch to another resource through the Round Robin DNS server. Thus, in a happy day scenario, both cars fall in and distribute traffic across the array. When one of these machines is turned off, the Round Robin DNS server, combined with retrying the browser browser, should ensure that browsers transfer their target host to a machine that is still running when they press the timeout. This is not what I came up with, but it seems to be a very good solution.
The problem I am facing is the following. The offset does occur, not only once for a failed request, but for each subsequent request from the same browser. Thus, a simple page request requires 21 seconds, after which all images also load in 21 seconds. The entire next page request also takes a lot of time. Thus, crash recovery works, but at the same time is completely useless.
Yield from digging:
; <<<> DiG 9.6.1-P2 <<→ example.com
;; global parameters: + cmd
;; Got a response:
;; →> HEADER <- opcode: QUERY, status: NOERROR, id: 45224
;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, OPTIONAL: 0
;; SECTION QUESTION:
; Example.com. In a
;; :
www.example.com. 86400 IN A 1.2.3.4
www.example.com. 86400 IN A 1.2.3.4
;; : 31
;; : 172.16.0.23 # 53 (172.16.0.23)
;; : 20 12:21:25 2010
;; MSG SIZE rcvd: 67
!
Kingsquare