Every single layer has to agree upon what it's looking at from the input (JSON is always UTF-8) to the parsing to the writing to the DB to the fetching from the DB to the UI. So if the characters are being corrupted, it might be because your DB is actually expecting EUC-KR or something. In that case you might try encode("EUC-KR", $utf8_string); check your DB schema and see what it expects. But the problem might be in the UI uptake or output or your DB might be incorrectly configured/specified.

Update: didn't see your reply before I replied. I might look at this later but it's a lot to digest. :(

So I have my db accepting these korean characters and can see them in korean in the tables too. I cannot use EUC-KR as I might have to accept all kinds of languages as input (one byte/two byte)

Below are the methods that deal with the encoding/decoding So I'm calling handleRequest($fun,$parameters) from my cgi file which calls a Handler.pm which initially does an eval call to handleRequest which firstly does eval call looks like this:

sub handle_req_cgi{
eval {
       $log->info("pjb.cgi - Started function $function , user: $user_
+name");
       $msg = UI::PJB::handleRequest($function, $parameters);
         1; 
    }

//Response
$tx->res->code($res_code);
$tx->res->headers->content_type('application/json');
$tx->res->body(encode('UTF-8',$msg));
#THIS IS THE LINE which corrupts the msg but if I print just the msg w
+ithout encode then #wide character warning is thrown by Mojo
$tx->rendered;
}
[download]

handleRequest() 
# looks like this:
{
 $jsonDecodedParameters = $J->decode($parameters);

my $result = executeFunction($fun, $jsonDecodedParameters);

if(ref $result eq ref {} && defined $result->{$HTTP_REQUEST_REDIRECT_U
+RL_KEY}) {
        $jsonEncodedResult = $J->encode($result);

 returns this $jsonEncodedResult 
}
[download]

then the control inside Handler.pm proceeds to write the response of

Maybe there's something I'm not understanding, but there's nothing in the first snippet that calls the "handle_pjb_cgi" function in the second snippet, and so I have no idea what the "$tx" thing is. Assuming "handle_pjb_cgi" is being passed the sort of object it should be getting, the description in the OP makes it sound like whatever file handle is being used to "print" the the "message" has not been set to use a utf8 layer.

I gather from your other reply below that you've confirmed there is valid utf8 data stored in the database, so the problem would seem to be limited to the output process, and probably involves the discipline setting on the output file handle. But I'm just guessing here.