Hide Forgot
Created attachment 1202541 [details] evm.log Description of problem: CFME 5.7.0.0 Steps to Reproduce: 1.In CFME: Compute ---> Containers --> Projects Notice how each object appears twice in the table 2.To verify that every object appears twice: Compute ---> Containers --> Routes Compute ---> Containers --> Container Services Compute ---> Containers --> Replicators Compute ---> Containers --> Pods Compute ---> Containers --> Containers Compute ---> Containers --> Container Nodes Compute ---> Containers --> Container Builds Compute ---> Containers --> Image Registries Compute ---> Containers --> Container Images 3.In order to get the actual number of objects: Computer ---> Containers ---> Providers 4. Click on Provider 5. Check the actual number of objects in the Relationships table Actual results: When accessing objects through Computer ---> Containers ---> Pods(Nodes, etc), each object is displayed twice in the table. Expected results: Each object should only be displayed once.
Created attachment 1202542 [details] audit.log
Created attachment 1202543 [details] production.log
Created attachment 1202545 [details] pods are displayed twice
Created attachment 1202547 [details] actual number of pods
This might be a clone of https://bugzilla.redhat.com/show_bug.cgi?id=1369359 Tony did you delete and re-add the provider? Are you running the latest build?
(In reply to Ari Zellner from comment #7) > This might be a clone of https://bugzilla.redhat.com/show_bug.cgi?id=1369359 > Tony did you delete and re-add the provider? > Are you running the latest build? Ari, I had deleted and added back the provider, but when I did it in CFME 5.6, it did not have any effect on the number of objects appearing in the UI. I am running the latest build - CFME 5.7.0.0 The issue is very similar to https://bugzilla.redhat.com/show_bug.cgi?id=1369359 but in this case, we are only dealing with one provider as opposed to two, also I had waited a while before adding the provider back, just to make sure the all objects were deleted.
So looks like this is caused by an AutomationEngine bug. Some events trigger a provider refresh and looks like they trigger a refresh even after the provider was deleted. So all of the providers entities suddenly re-appear having entity.ext_management_system = nil. The 'leftover' entities arent leftover, they are new entities that were created after the provider was deleted. Disabling the automation engine stops this from occurring.
(In reply to Ari Zellner from comment #9) > So looks like this is caused by an AutomationEngine bug. > Some events trigger a provider refresh and looks like they trigger a refresh > even after the provider was deleted. So all of the providers entities > suddenly re-appear having entity.ext_management_system = nil. > > The 'leftover' entities arent leftover, they are new entities that were > created after the provider was deleted. > Disabling the automation engine stops this from occurring. Is automation engine enabled by default?
It was enabled on the env you gave me. You can test this out by creating->refreshing->deleting providers with the automation engine on and off. Youll see that the number of entities continues to increase if the automation engine is enabled.
I'll test it with the automation engine on and off, I wonder though, when an appliance is created and no providers were added yet, is the automation engine supposed to be on or off? I think the default state is on, right?
Good question, I think its customer specific.
This is not a blocker , it looks like a specific item that comes up only in automation. The has a lesser priority for GA, Hence removing the blocker flag & moving to zstream
Possible solution discussed with core team: https://github.com/ManageIQ/manageiq/pull/13204
Einat does it happen in automation ? and how often ?
Indeed this happens in automation. However, it happens intermittently. One of the test cases we have deletes and adds a provider immediately causing the appliance to, sometimes, show objects with no provider attached. We are working on fixing this test case, however, dev should look into why objects are not being totally cleaned from the DB once a provider is deleted.
Federico, Our CI has been refactored and the issue has been resolved. This was happening since in the automation scripts, a provider was deleted and added immediately i.e. CFME did not have a chance to "recover". As discussed, this is not common practice at a customer site. QE will investigate further, however, for right now, this issue is not holding up automation any longer.
OK closing for now because we don't have any way to reproduce or enough information. Feel free to reopen if relevant.