incorrect 4-Byte UTF-8 conversion

[Alhimika]

I know this sounds a bit silly without providing details, but something is not right with MyBB internal 4-Byte UTF-8 conversion.
At first no no conversion mistakes but when trying to enter unicode smiles into posts perhaps on 10% getting stored and parsed correctly. 

This is a something I tried with 2 different hosters - same thing 

Same conversion works ok with a test forum I've set up not with my actual forum. 

Honestly no idea from where to start

[frostschutz]

This user has been denied support.

encoding set correctly in the database (all tables and fields)?

database encoding set correctly in inc/config.php?

[Alhimika]

database encoding set correctly in inc/config.php?

Yes it is. I've set it to 

$config['database']['encoding'] = 'utf8mb4';

encoding set correctly in the database (all tables and fields)?

Sorry my DB knowledge is very limited
I can go to phpmyadmin, but please tell me what should I look for 

Thanks

[Omar G.]

Does your server meet the minimum MySQL 5.5.3 requirement?

[Alhimika]

Does your server meet the minimum MySQL 5.5.3 requirement?

Hi! yes it does

BTW same server I have a test forum - just for trying different thigs before implementing on my main forum
Conversion goes ok over there 
https://alhimika.net/forconsole/

same server, same setup, same actions, but with above mistakes 
https://alhimika.net/forum

[Alhimika]

Anyone?
Or perhaps can you recommend some software tools to work with DB to check it and to make end user life a bit easier.

[Omar G.]

Hi, so you ran the MyBB tool to convert tables, so I assume the end result is a all-green checks?

If so, what happens when you edit a post that you found is not parsing unicode smiles?

[Alhimika]

Hi, so you ran the MyBB tool to convert tables, so I assume the end result is a all-green checks?

If so, what happens when you edit a post that you found is not parsing unicode smiles?

Yes end result is all green

after conversion about 10% getting parsed as they should
But probably 90% parsed into question marks, same as on 3 byte utf 8
Obviously if without conversion all of them converted into question marks
At the moment I reverted back to original UTF 3 byte using backup. 

Maybe, just maybe there something broke in my DB? 
It's quite an old and originally was converted from VB3 

As I mentioned earlier if I do same conversion on my test forum, no problems whatsoever

[Omar G.]

Oh, I remember I once did a conversion and indeed unicode smiles get converted to questions mark, I think this is the expected result? Maybe somebody can jump into the discussion.

[frostschutz]

This user has been denied support.

conversion can not fix already broken characters in old posts

but it should not break additional characters on its own accord

if you are not converting from utf8 to utf8mb4 but from something else (latin1) there may be complications (if utf8 characters got stored in latin1, which is weird)