What are characteristics of a "good" thesis topic?
I completely agree with ff524's criteria, but here are some further considerations:
A good thesis topic should not be all or nothing. Instead, it should accommodate a range of levels of ambition. There should be fallback options that would be enough to graduate even if things go poorly, moderately ambitious outcomes that you could reasonably aim for, and very ambitious outcomes that would be fantastic if they could be achieved.
It should offer scope for creativity and choice by the student. You can only impress hiring committees so much if it looks like your advisor was the mastermind behind your thesis. Instead, it has to be clear that over time you took ownership of the project and played a major role in determining its direction (even if your advisor made the initial suggestion and provided feedback and guidance).
Your thesis topic should connect with other topics you'd like to explore in the future. If you spend your entire early career revisiting and elaborating on your thesis topic, it won't set you up for success as a researcher. Instead, you need to branch out, and your choice of thesis topic can really help with this. Choose a topic that is central and well-connected to adjacent areas, rather than one that is isolated. That way you can build on your previous experience as you branch out.
Your thesis advisor needs to care about the topic. Sometimes people are willing to supervise theses on topics they don't particularly care about. They do this in an attempt to be helpful, but it can end up working against you when you graduate and discover that your advisor's lack of caring translates into a lack of support on the job market.
For a multi-year project like a thesis, it's generally best to pick a topic that seems to be becoming more popular and influential over time, rather than less so. This sounds obvious, but it's surprisingly subtle to figure out. Beginners are sometimes a little behind the times and can get excited by topics that were hot a few years ago but have since been thoroughly investigated and are already in decline.
It should be something you like doing, not just something you would like to have done.
A "good" thesis topic is one that's
- a good research topic in general (moves the field forward, publishable, etc.),
- interesting to you specifically (you'll be spending a LOT of time on it),
- a topic that (you and your advisor believe with some degree of confidence) can be addressed with the resources that are available to complete the thesis.
("Resources" including but not limited to: time, money, lab equipment and materials, supervision, etc.)
The answers by ff524 and Anonymous Mathematician have great advice on how to evaluate a thesis topic.
As another approach, start by considering advisors that you think you may like to work with. Don't underestimate how important it is to find the right advisor. This may help you to explore what they're interested in and how it overlaps with what you're interested in.
And once you have an idea about the advisor and a topic, reach out to them! They should be more than happy to discuss potential projects with you and this will definitely help in your search for a topic that you really want to do!
Finally, I have to object to this concern in your question:
I'm sure some will be solved before I start my actual thesis
Problems in academia are rarely 'solved'. If you find a topic that you think has been solved, speak with the advisor for the project and I'm sure that they will already be considering ways to extend the topic. There is always more work to be done! Sometimes it's better not to start from scratch, although this may be discipline specific.