Bug #1186
closed
Report: No answer and directive with property "unique"
Added by François ARMAND over 13 years ago.
Updated about 10 years ago.
Category:
Web - Config management
Description
When two directives from a technique with the "unique" property set to true are applied to the same node, and one directive has a lower priority than the other, the one with the lowest priority is reported to have status "no answer".
It's "normal", has that directive is in fact not applied (because it's the one with the highest priority which is), but the user should have that information in reports, and not a misleading "no answer" one.
Moreover, the fact that that report is "no answer" may lead to incorrect aggregated report status for the rules.
C'est une remarque très pertinente.
Néanmois, ca n'est pas une info que je stock dans les rapports attendus. Il y a deux approches :
- changer les rapports attendu pour prendre en compte les PI effectivement appliqués
- filtrer a chaque affichage celle qui sont appliquées
- utiliser le contenu de NodeConfiguration pour savoir ce qu'on attend sur une NodeDetail
Je pense que la 1er est la plus pertinente
- Category changed from 6 to Web - Config management
- Target version changed from 9 to 10
- Status changed from New to Discussion
- Target version changed from 10 to 19
Quelque chose à voir avant de sortir la 2.3.0 Finale ?
- Target version changed from 19 to 21
- Target version changed from 21 to 23
- Target version changed from 23 to 18
- Target version changed from 18 to 24
- Target version changed from 24 to 18
- Subject changed from Report: No answer et policy unique to Report: No answer and directive with property "unique"
- Description updated (diff)
- Status changed from Discussion to 8
- Target version changed from 18 to 2.4.0~rc1
- Priority changed from N/A to 2
It seems to be a rather important bug, as if I understand correctly, any rules node with at least two "unique" directive from the same Technique will have wrong reports.
This is indeed a problem, but I'm afraid the impact of this modification for the 2.4 would be too big. When we compute the expected reports, we do it on a rule basis, and not on a node basis. Changing that would imply massive modifications, with a risk of malfunction.
Plus, it would be important to show to the user that a directive has been overriden by another, otherwise he might have unexpected behaviour.
The current solution with no answer shows that something happens, which is better than nothing from my point of view
- Status changed from 8 to New
- Target version changed from 2.4.0~rc1 to 24
Nicolas CHARLES wrote:
This is indeed a problem, but I'm afraid the impact of this modification for the 2.4 would be too big. When we compute the expected reports, we do it on a rule basis, and not on a node basis. Changing that would imply massive modifications, with a risk of malfunction.
Plus, it would be important to show to the user that a directive has been overriden by another, otherwise he might have unexpected behaviour.
The current solution with no answer shows that something happens, which is better than nothing from my point of view
I agree that this would be too big of a change for 2.4 at this stage. Also, it must be noted that there is currently only have one Technique that is "UNIQUE" (MULTIINSTANCE=false), and it's motdConfiguration - not really mission critical.
Let's move this to 2.5.
- Target version changed from 24 to Ideas (not version specific)
- Status changed from New to Rejected
Also available in: Atom
PDF
Updated by 
Updated by Jonathan CLARKE 
Updated by