0

In a cluster (say a cluster in case of application servers), what happens to a request which has already reached the server, which has recently failed. Does it get lost or is it retransmissioned from the admin server to the nearest node? or anything else?

Gaurav
  • 1,570
  • 4
  • 20
  • 25

1 Answers1

1

I would say that depends on the kind of cluster and it's configuration. You can't say that for all clusters but most scalable systems are designed to have no single point of failure and be as state-less as possible. Therefore the request gets lost and is not executed. The requesting client is responsible for restarting the request, which may then be dispatched to an other node.

Tobi
  • 540
  • 7
  • 18
  • Thanks for your answer, is it also possible that the Admin Server(which manages the managed servers) keeps a copy of the request and after knowing that a node has failed, it can send the copy of request to the nearest node (though there might be an overhead of keeping a copy of every request on Admin server) – Gaurav Oct 22 '13 at 06:22
  • Generally it is possible, depending how the cluster is designed. – Tobi Oct 22 '13 at 09:25