Roberto Polli

### Reproduce - I have 2 infra nodes ``` infra-0 infra-1 ``` - infra-1 fails, I mark it unhealty and update the stack ``` o stack resource mark unhealthy shift...infra......

Note: the following procedure just fix the heat part of node recovery. There is another issue with ```fragments/bastion-node-cleanup.sh```, that is run against the new node: I will fix this in...

### For OCP3.5 As for https://bugzilla.redhat.com/show_bug.cgi?id=1458660 in case we wanto to set openshift_http_proxy, we need to add the fixed_network to openshift_no_proxy

### I expect on openshift >= 3.4 fluentd port is not needed and not open by iptables ## Instead openshift.yaml opens 24224 tcp/udp ## Further infos https://bugzilla.redhat.com/show_bug.cgi?id=1386329

masters are exposed on floatingip

1

# I expect there should be an option to avoid exposing masters. From the doc: ``` Master nodes and infra nodes have floating IP assigned to make sure these nodes...

### I expect that using registry_ephemeral don't create a persistent volume ### instead the content is the same of registry_persistent

### Description openshift nodes are log-spammed with ``` May 26 05:20:59 shift-33-node-4o2c2g93 kernel: intel_rapl: no valid rapl domains found in package 0 ``` ### We could either - disable rapl...

A general way to configure openshift vm in a single place. - Define an OS::Heat::MultipartMime in openshift.yaml - Pass it to all hosts For now it just contains a Timezone...

The `extra_openshift_ansible_params` requires valid json, while we could easily add yaml configuration and comments to our files, eg. All the issues (syntax, override,...) that are still present with `extra_openshift_ansible_params` apply....

### I expect that changing rhn_* configurations won't recreate the bastion node. When using static root volumes, this could cause the stack to fall in an unrecoverable state ### instead...