Myself and Logan (especially Logan, huge kudos to him for working so hard on this) have spent countless hours trying to figure out WPMS.
After much investigation, writing custom code, making unreasonable amounts of changes to things that should work out of the box, we’re close to concluding that Wordpress Multisite is not the right implementation of Wordpress.
It just doesn’t work, here’s why:
Preview doesn’t work on multisite with domain mapping
Random redirect loops
Custom code to make SSL work (we’re an operations team, we shouldn’t be maintaining a code base)
Would like to see some documentation the planned architecture for this Wordpress-as-a-Service (WaaS) for us to review at the next team meeting. I don’t want to just rubber stamp this without some input from a wider audience.
How will instances be managed?
How will this WaaS - instances & database - scale or autoscale?
How do we ensure S3 storage for media content/themes?
I always like to lead with requirements but as I tried to write them down, I came up with only two high level requirements:
@logan & @tad - as module owners, you should take the lead at a draft proposal that we can discuss.
(Though this Sunday I’ll be back on a plane from Chicago to San Francisco - perhaps we can schedule it for some other time when the right people can join?)
Well, the problem isn’t that the plugins are bad, it’s just that they aren’t comprehensive and nobody keeps them up to date.
Logan basically rewrote a whole HTTPS plugin which got us further but not where we needed to be.
The redirect loops are very random on login pages, and we can’t reproduce it frequently enough to debug.
WordFence doesn’t break stuff.
Right now I’m experimenting with container based wordpress using docker, managed by tutum which seems to be a very good solution. It works well, gives us a really high level of security and easy to load balance.