You should remember that the rating assigned to problems is estimated from an aggregate statistic, and not individual submissions — individual contributions to the difficulty estimate become less important as more and more people solve a problem.

Meanwhile, your perception of difficulty is a single datapoint, which has much higher variance. A lot of it is also dependent on whether you "see" the solution immediately or not. And we can't discount the fact that people learn different topics at different rates.

So don't think too much about the difficulty being correct or not — focus on solving problems outside your comfort zone. A good way is to go into the problem without knowing the topic and after you solve it, look at the tags and the rating to get an idea of what your strengths and weaknesses are. Then figure out whether it was a knowledge issue or a skill issue, and work on it accordingly.

CF ratings feel a bit hit-or-miss in the lower ends (or maybe it's just my own bias), but statistically they're not unusable. If you want other rating systems, clist.by has some problem ratings that you can look at and judge for yourself. I haven't used any rating system for CF problems other than the CF system itself personally.

I wasn't able to solve it when I first tried. ;_;

so why you solve it 2 days?

Different people are good at different types of problems. I got stuck on that problem during the contest for 30 minutes and decided to do E first before I saw the solution.