Hi Georges.
Do you do any scripted restarts on a cadence today? As an On-Premise Windows infrastructure admin, I can sing the praises of scheduled tasks. There's just too much going on in your infrastructure that you will never know about and will never be in any vendor's test infrastructure & documentation, nor may ever be resolved by a support case.
I do a weekly scheduled restart of all services on all servers (service stop all, service start all), and run a service start all every 5 minutes. I implemented this over 6 years ago and this makes all kinds of these kinds of weirdnesses just go away. I highly recommend it.
By all means open a support ticket and see if CA can ID something amiss with your configuration - but if not, take action to set yourself up for success in your organization's infrastructure, which is outside of CA's ability to influence.