Strange Redirect Behavior

Couple of things come to mind.

First, as a best practice, you should include a "--" in your switch statement. This tells the switch command that there are no more arguments, so someone construct a request URI starting with a "-" character and break your rule.

Second, your switch is not using a "-glob" argument. If you don't have a default pool on the VIP and the request URI doesn't exactly match "/devicepairingstage" or "/devicepairing", the LTM has nowhere to send the traffic.

Perhaps you should try something like this:

timing on
when HTTP_REQUEST {
  switch -glob -- [string tolower [HTTP::path]] {
    "/devicepairingstage*" {
      pool ianappsstage-pool-https
      log local0.debug  "uri: [HTTP::uri]"
      log local0.debug  "[client_addr] hit devicepairingstage"
    }
    "/devicepairing*" { 
      pool ianapps-pool-https
      log local0.debug "uri: [HTTP::uri]"
    }
  }
}