----- "Scott M Ferguson" <smferguson(a)gmail.com> wrote:
Hey all. I have a rough proof-of-concept going for the client/server,
but in chatting it over with a friend a novel idea came up. What if
the server just opened 2 ssh sessions for each client. 1 to run the
tests, 1 to tail the log (grab the result). Then we wouldn't need a
client daemon. I'm not sure this is desirable since other projects
use
a standard client/server model, but thought it was worth bringing up.
Best,
Scott
Hello Scott,
I'm not exactly sure I get the idea. The autotest server already
handles starting the task and collecting results (arrows 2. and 3. in
the picture I added to
https://fedorahosted.org/autoqa/ticket/183#comment:description). Ticket
#183 mainly concerns arrow 4., signaling to the host (of that autotest
client VM). We can use it for reverting the VM to the previous state
(I think that's the main reason we need all of this). In the Virtualization
milestone there are other tickets that cover remaining parts of that
picture.
So, the problem of ticket #183 is not getting the results (autotest handles
that for us), but telling the host (of that VM) "do something" right after
completing the test.
Are we on the same page? Maybe I have missed something. Tell me.
Thanks,
Kamil