On 14/02/14 15:27, Silvio Moioli wrote:
On 02/14/2014 01:05 PM, Cliff Perry wrote:
http://turing.suse.de/~smoioli/action-chaining-mockups/list.html
I suspect this page really wants a column for when is the event
scheduled to be executed.
Not really - those Action Chains are not scheduled for execution yet.
When they actually get scheduled, those become regular actions in
Schedule -> Pending Actions (and later Completed Actions, Failed
Actions, etc.).
OK :)
Merge/use either only created or modified, vs both - I don't think we
need them both.
Okay, agreed. Task scheduled for next sprint.
http://turing.suse.de/~smoioli/action-chaining-mockups/editor.html
This is a bit worrying, it looks like it would make it hard/complex to
render information meaningfully. Because they are chained and you have
different systems for each portion of the event - does it mean, you
don't reboot the 10 systems, if one of the two config deployment fails?
No, the chaining is done per-system, so if an action fails on a
particular system then only subsequent actions in the same Chain for the
same system will fail.
OK.
See:
http://wiki.novell.com/index.php/SUSE_Manager/ActionChaining#Definitions
Rendering basically works at the moment, I still have to fix a couple of
bugs but it looks very similar to that mockup. I already planned to
prepare a screencast for you during the next sprint so that you can also
see it.
While the power to add/remove systems for each portion sounds nice, it
makes my brain hurt, and I'd have to assume it would for those managing
the systems - selecting X systems and being stuck to work with them
only, sounds more sane approach.
We do expect that users will mostly use the SSM with a fixed set of
systems to chain actions, and this will be the recommended use case anyway.
Indeed :)
Please note that using the SSM automatically implies having chains with
actions that are executed on different sets of systems. This is because,
in general, you do not get one action per system that you have in the
set as some of them might not meet the prerequisites (entitlements,
channels, etc.).
So, from a UI perspective, we could not really do much more than reflect
that fact, so we came up with that "tree-like" visualization scheme.
From there, adding the capability of deleting a system from an action
was just a low-hanging fruit, it could also be disabled easily if you
feel that it would hurt usability.
Adding a system to an existing action in a chain is not possible at the
moment.
I don't see anything bad in getting this into Spacewalk, or that would
go against Spacewalk's future desires.
Great news!
I will be back with a screencast soon.
Great. And sorry for delay in my own review of this :)
Cliff
Thanks a lot,
_______________________________________________
Spacewalk-devel mailing list
Spacewalk-devel@redhat.com
https://www.redhat.com/mailman/listinfo/spacewalk-devel
