I strongly agree.   Start by using a debugger in your web-browser to definitively observe what actually is being passed between the host and your client.   Don’t “assume” anything.   There are just too many places (on both sides) where a problem could be cropping up, such that the otherwise-observable symptoms of the problem really don’t point to it at all.   A debugger will quickly remove all doubts:   “Here is what the host sent, and here is what the client POSTed back (and it was a POST ... don’t assume ...), and here is what the host returned.

Then, as necessary, insert logging-messages into your Perl (server-side) code, so that (over and above what the web server already records ...) it also documents exactly what it is seeing and doing.   I use server-side debug messages a lot ... both “verbose” messages (which can be turned-off unless needed), in which the server positively “blabs” about what a wonderful day it’s having today, and “warning” messages when something appears to be wrong or questionable.

Once you can clearly see the (often, well hidden) evidence of a problem, its solution is usually obvious and straightforward.   Aye, there’s the rub ...