Not a standard problem with Russian characters. How to fix encoding?

Question

Character encoding

The problem is that when the page is updated, the encoding of Russian characters in the string changes automatically and randomly, from the normal "ytsu" to "������"
Absolutely all site files are in utf-8, the base too.
Help to understand the reason for this magic.

Answer 1

<meta charset="utf-8">
commented out

Answer 2

try the Text decoder together with the result of the restored source text will give possible variants of the correct encodings. Only if you analyze by downloading a file, remove the excess - the decoder cuts the input text to 100 characters