I expect this is complaining about the old PHP in EL7. The version in EL7 is 5.4 which is no longer supported by newer versions of Wordpress. The fix would be one of the following:

1. Move communityblog into the wordpress cloud and have them manage it.
2. Move communityblog into a F30 container which would sit in the next generation community openshift cloud.

Either one of them will take a project plan and time to do and would probably be 2-3 months out.

If we can move it to WordPress and still use FAS for auth, it seems like option 1 is the best (because it's one less thing for the infra team to manage). In either case, a 2-3 month wait would be fine.

I'll run this by the rest of the CommOps team to see if there are any strong opinions and come back to you.

FAS login is currently implemented by using the Wordpress OpenID plugin, together with a custom extension to automatically pre-fill the OpenID identity and submit the login form which is available at my Fedora People page.
It might be good to look at moving to OpenID Connect at some point, but that's an entirely different problem :).

I've updated the issue to reflect the current state. @jasonbrooks got us an instance on OSPO's WP Engine account, so we're going to migrate the Community Blog to that. We'll need to coordinate to do a final sync of the content and change the DNS entry. CommBlog editors are testing the new instance now, so I'm thinking we'll target sometime during the last two weeks of July to make the move.

@bcotton How is auth set up on this new instance? Using OpenID, OpenID Connect (unlikely since I've not heard requests for a Client ID?), or local users?

How is auth set up on this new instance? Using OpenID, OpenID Connect (unlikely since I've not heard requests for a Client ID?), or local users?

@puiterwijk it's using your OpenID plugin and extension like the existing site is using.

Who can @jasonbrooks and I schedule the DNS update with?

@bcotton I can do it.

I would recommend to change the haproxy backend rather than DNS ( as a first step) . For a start, DNS is not immediate to propagate, and since the wpengine backend use lets encrypt, we would need to wait a bit for the ssl certificate to be ready. In turn, this mean people may see a ssl error warning, which is kinda annoying since people will have to say "no, that's normal, that's temporary". Plus, a proxy change is easier and faster to revert IMHO, in case something go wrong.

Ok what is the status of this at the moment?

Apologies. We changed the proxy as misc suggested. This is now complete. I'll open another issue when we're ready to shut down the old instance.