SoftwareDeploymentGroup does not accept any attribute from ResouceGroup for "servers" property

Bug #1544728 reported by Christopher Hultin
20
This bug affects 4 people
Affects Status Importance Assigned to Milestone
OpenStack Heat
Confirmed
Low
Randall Burt

Bug Description

OS::Heat::SoftwareDeploymentGroup requires a Map containing the server IDs/names to apply the SoftwareConfig to. OS::Heat::ResourceGroup returns a list of server IDs, rather than a map. There are no intrensic functions that permit the conversion of a List to a Map, preventing the usage of a ResourceGroup and a SoftwareDeploymentGroup.

Revision history for this message
Christopher Hultin (chris-hultin) wrote :
Revision history for this message
Randall Burt (randall-burt) wrote :

Steve Baker: maybe I'm missing something, but I can't find a good way to match these two up though it should be the easiest thing. Thoughts?

Revision history for this message
Steve Baker (steve-stevebaker) wrote :
Changed in heat:
status: New → Invalid
Revision history for this message
Randall Burt (randall-burt) wrote :

randallburt stevebaker: though that seems to be a lot of hoop jumping for two things that should "just work" together. Would there be any objection to adding another property to SDG that accepts a list of server ids? The names of the resources in the SDC nested stack don't seem terribly important on the surface.
3:23 stevebaker randallburt: it is important, the key in the map determines the name of the deployment resources
3:24 randallburt stevebaker: but why should we care if its some provided name or an index?
3:25 stevebaker randallburt: because servers may come and go from the list, so an index may end up refering to different servers - which would be bad
3:26 stevebaker randallburt: this way we can use ResourceGroup's behaviour of not reusing names (which are incrementing numbers)
3:26 randallburt stevebaker: so I'm assuming here that I *have* to use a nested stack that simply redefines the server resource, but I could just use a server?
3:27 stevebaker randallburt: sure you could define your own servers map with arbitrary keys
3:27 randallburt stevebaker: I'm talking about using a ResourceGroup that has vanilla OS:::Nova::Servers in it
3:28 Drago get_attr: [rg, attributes, accessIPv4] ?
3:28 stevebaker oh, yeah. that should work
3:28 randallburt Drago: well, not the ip address. SDG needs server ids
3:28 Drago randallburt: ah right
3:30 randallburt stevebaker: so OS::Nova::Server doesn't have an attribute for its id, how would that work, then?
3:31 dprince left the room (quit: Quit: leaving).
3:32 randallburt stevebaker: seems like we would have to change https://github.com/openstack/heat/blob/master/heat/engine/resources/openstack/heat/resource_group.py#L420-L422 to return refs if no path were specified or something similar
3:33 stevebaker randallburt: I suspect its already possible, just poorly documented
3:33 Drago get_attr: [rg, attributes, show, id] ?
3:34 stevebaker randallburt: don't you want get_attr: [rg, refs] ?
3:34 randallburt stevebaker: yep, but that's not a map, is it?
3:35 stevebaker randallburt: heh, is there a problem with putting the server in a nested stack?
3:35 stevebaker Drago: that would work but would hit nova for every server
3:36 Drago oh
3:36 randallburt stevebaker: seems hoop-jumpy and really unnessisary. I'd define a template thats just a wrapper around OS::Nova::Server just to get this to work...
3:36 randallburt stevebaker: and again, it seems odd to me that these two things don't "just work"
3:36 randallburt together
3:37 stevebaker randallburt: sure, but I'd rather rg be enhanced to create a refs map than sdg be enhanced to take a list
3:38 randallburt stevebaker: that works too
3:38 vijendar left the room (quit: Quit: Leaving.).
3:38 randallburt stevebaker: so copy pasta this convo into the bug and re-open. I can hack the change into RG.
3:39 randallburt stevebaker: that was a question, btw. I type dumb.
3:39 stevebaker randallburt: +1
3:39 randallburt stevebaker: cool, thanks!

Changed in heat:
status: Invalid → Confirmed
importance: Undecided → Low
assignee: nobody → Randall Burt (randall-burt)
milestone: none → mitaka-3
Changed in heat:
milestone: mitaka-3 → mitaka-rc1
Changed in heat:
milestone: mitaka-rc1 → newton-1
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.