Loading...

Type: Bug
Resolution: Unresolved
Priority: Major
Fix Version/s: quay-v3.13.2
Affects Version/s: quay-v3.9.2
Component/s: quay-operator
Labels:
- openshift
- quay
- triaged

Blocked:
False
Blocked Reason:
None
Ready:
False
Affects:

Documentation (Ref Guide, User Guide, etc.), Compatibility/Configuration, User Experience
Steps to Reproduce:
Hide

In order to reproduce the bug:

Edit the Quay CRD and set the HPA as "managed: false"

Edit the HPA resource and increase the value of minReplicas field

In this case, the default number of replicas for the deployment is 2, so try to increase the number of minReplicas in the HPA to a higher value (e.g.: 4).

An override of the number of replicas for the Quay deployment in the Quay CRD can be also added, as mentioned in the bug description.
Show
In order to reproduce the bug: Edit the Quay CRD and set the HPA as "managed: false" Edit the HPA resource and increase the value of minReplicas field In this case, the default number of replicas for the deployment is 2, so try to increase the number of minReplicas in the HPA to a higher value (e.g.: 4). An override of the number of replicas for the Quay deployment in the Quay CRD can be also added, as mentioned in the bug description.
Intelligence Requested:
Market:

Target Version:

quay-v3.13.2

SFDC Cases Links:
SFDC Cases Counter:
SFDC Cases Open:

PX Priority Data:
PX Impact Score:

Editing the Quay CRD in order to set HPA as unmanaged and then increase the number of minReplicas in the HPA resources causes a conflict between Operator and HPA in the management of replicas.

As a consequence, new pods are created by the HPA but removed by the Operator continuously, so the HPA stops working.

This behaviour can be observed in the following scenarios:

HPA set as unmanaged in Quay CRD, keeping the deployment as managed, and minReplicas of HPA increased
HPA set as unmanaged in Quay CRD, keeping the deployment as managed with an override to increase the number of replicas, and minReplicas of HPA increased to the same value
- In this case, continuous pods creation and deletion start when HPA tries to scale them

It seems there is not a way to increase pod replicas while keeping the HPA properly scaling them.