Many people have had the experience of choosing a lightweight framework (Sinatra, Flask, etc.) and gradually needing more and more features from a heavier framework. Eventually you find that you wrote a poorly documented, poorly implemented version of Django or Rails, and bringing new people onto your team is now far harder than just finding people who've used a specific framework.
Then I suggest you scratch both the "popular" and "similar to Ruby and Rails" requirements. When you do that, you have things available to you such as Elm and Reason (with ReasonReact) for the frontend and F#/Haskell/Scala on the backend.
Popular will tend to the lowest common denominator, and similar to what you already know means you won't gain much if at all :)
The great thing with Yii is performance (efficient generated SQL queries, multilevel caching, mixed eager/lazy loading) and instrumentation for application tuning. Other stuff is like other good frameworks.
It has a great community too, lots of great packages(gems).
It might be more interesting to ask "what is nothing like Ruby and Rails?" to push yourself to learn something new that's entirely different, can give you new feelings.
If you are interested in exploring truly new things, there's a lot of interesting stuff happening in Universal or Isomorphic JS space (both names are basically for the same thing, and about equally common in usage), some of which becomes very different from the traditional Rails approach to a web backend. (Also, there's a lot of interesting variety in language options from transpiling ES2018 or ESNext to Typescript to increasingly more obscure transpiles to JS languages. I recommend Typescript as the best place to be, for what that is worth, but you'll get a bunch of other opinions pretty easily.) It could be useful experiencing some of that, and get a very different experience from just "Rails but in a another language". Server-side React seems to be getting increasingly popular, and the GraphQL approach to database work can be very different from a traditional ORM approach like ActiveRecord.
Phoenix is similar enough to Rails that you won't feel totally lost, while using Elixir will teach you about functional programming and actor systems.
Lucky uses crystal, you get all the benefits you are used to with rails, i.e. migrations, models views etc.
The main benefit is that crystal is a compiled language and is type safe. So you get to use all your ruby skills and the compiler adds a whole new level of safety to your code.
Elixir/Phoenix is far less popular, but very fun and gaining steam.
I’d pick Node JS given that most web devs end up having to use Node anyway even if just as a build pipeline. Knowing Node better will always be useful. Also the dynamic typing will feel at home for a Ruby dev, and if you want to use types then try Typescript. If you want to go functional there is Purescript, Reason etc. All compile to JS so you can use on Node.
Many huge sites are running on rails though, so switching to a new framework isn't going to give you automatic "scale". I'd recommend reading this post which summarizes a re:Invent presentation about scaling to 10 million users on AWS http://highscalability.com/blog/2016/1/11/a-beginners-guide-...
But i really don't see the benefit of learning that stack when you already know Ruby+Rails. They are veryyy similar and unless your absolutely need some special Python library, it is simply a matter of personal preference.
It seems a pretty great framework similar to rails could be assembled out of typescript, next.js, react, typeorm, webpack, ant design, terraform, postgres, cloudflare workers, cloudflare key value store, rds and by adding model and view generators, plus project and crud scaffolds.
Not so popular, but a number of those who have made the switch swear by it.