1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
|
I've got the test set up like so: A single Client (Supplier) with two
modes: NORMAL and FAULT_TOLERANT (FT). Every timeout, it does work then,
when in NORMAL mode, it pushes events to only the Middle1 process; when
in FT mode, it pushes events to both Middle processes. Currently, I have
it start in NORMAL and switch to FT after the first time the task
executes. There are 2 middle processes, Middle1 and Middle2. They simply
receive events, do work, and, based on the mode of the Client (they can
tell from the event type), push events in the same mode to the Server.
The Server subscribes to events from both Middle processes. If the
events indicate NORMAL mode, the Server performs work every time it
receives an event. If the events indicate FT mode, the Server performs
work only after it receives an FT event from both the Middle1 and the
Middle2 processes.
Thus, we have a single task with 3 subtasks: T1_1 is executed in the
Client. T1_2 is executed in Middle1 and if in FT mode in Middle2. T1_3
is executed in the Server when Middle1 is done (in NORMAL mode) or when
both Middle1 and Middle2 are done (in FT mode).
Here's how to run it: I have provided scripts (Client, Server, Middle1,
and Middle2) to run the processes. You need to run the Client first,
then the Middle processes (any order), then the Server. The Middle
processes run their ORB for 305 seconds and the Server runs its ORB for
310 seconds to accommodate the delay between starting each process.
Here's were to run it on the cluster:
Server (Ron)
Middle1 (Harry)
Middle2 (Hermione)
Client (Norbert)
|
