Bug 2024525

Hi all,

After upgrading RHCS 4 installation to RHCS 5 prometheus port for basic graphs is changed from 9092 to 9095.

The new port is reflected in our documentation [*], and there should be no issue with that for the new installations. But i believe that may introduce some challenges for the upgrade scenarios, since there for example can be configured the firewall rules, which would result in the service outage, or require the fw reconfiguration.

Trying to specify the old port number (9092) in service config yaml file results in a failure:

~~~
$ cat ~/cephadm-configuration/prometheus.yaml
service_type: prometheus
service_name: prometheus
placement:
  count: 1
  label: monitoring
spec:
  port: 9092

$ ceph orch apply -i ~/cephadm-configuration/prometheus.yaml
Error EINVAL: ServiceSpec: __init__() got an unexpected keyword argument 'port'
~~~

It looks we're hitting the following upstream bugs with that:

[adopted prometheus instance uses port 9095, regardless of original port number]
https://tracker.ceph.com/issues/45996

[cephadm: monitoring services adoption doesn't honor the container image]
https://tracker.ceph.com/issues/46561


Could there be implemented any workaround for preserving the port number in the upgrade scenario?

Thanks in advance,
Sergii


[*]
https://access.redhat.com/documentation/en-us/red_hat_ceph_storage/4/html-single/dashboard_guide/index#network-port-requirements_dash

https://access.redhat.com/documentation/en-us/red_hat_ceph_storage/5/html/dashboard_guide/ceph-dashboard-installation-and-access#network-port-requirements-for-ceph-dashboard_dash