I mentioned the background reposync on the main screen previously -- now
you can also select particular repos and run syncs on any combination of
them from the repo list page.
In addition, instead of being a long synchronous operation, power is
also now one background task... so you can power cycle 5000 systems and
then go do some real work while those finish up (not sure this use case
is crazy realistic unless you run an uber-cluster, but uber-cluster
folks, this is for you).
I think the logs look a little polluted ATM for the individual tasks,
logging some exceptions extra (Cobbler Exceptions get printed twice
because the catch for the CX raises a CX...) though we'll have this
cleaned up sooner or later.
All in all, pretty good.
There is a background_reposync XMLRPC api call as well as a
background_power call. The old power call still works for those systems
that were using it.
I think earlier Bill Peck mentioned something about the timing between
power operations being weird based on what systems were on the same
power management hardware. Given that, do we need to bump the timeouts
or do something more clever about when we sleep between operations? I
think now is the time but could use a review of what the right thing to
do is... (Added Bill on CC)
--Michael
Show replies by date